This is part of a series of helping founders understand VC incentives and dynamics. Subscribe to get the rest.

Subscribe now

Ownership Math

First let’s look at the basic math. A fund will spend around 20% on overhead with management fees (2% yearly for 10 years), and with the remaining 80% then decide to invest a share in follow-on funding, maybe 25%.

• First checks amount = Fund size * 80% investable * 75% first checks

• First check size = First checks amount / number of investments.

If you wanted 25 vs 50 first checks, for a $75M fund, that math would work out to:

• 25 checks: $75M * 80% * 75% / 25 = $1.8M

• 50 checks: $75M * 80% * 75% / 50 = $900K

This already makes clear that you should ask the firm you’re talking to about check size. If they want to typically invest $2M, that is an easier check to win than $1M. Demand curves in venture don’t slope downward.

Let’s assume only 20% earn a follow-on check, so respectively 5 and 10 checks. Similar math dictates the check size:

• 5 checks: $75M * 80% * 25% / 5 = $3M

• 10 checks: $75M * 80% * 25% / 10 = $1.5M

This is another area to ask about when you talk to a fund: do you follow on? How often? What criteria? How much?

You can see how this changes your concentration by assuming some typical round values. Let’s say the first check is in a seed round at $20M post and the follow on is in a Series A at $60M post. Fully doing this math requires paying attention to dilution and option pools, which I’ll do in a table here. To summarize the outcome, here is ownership after the Series A and after a few more successful rounds to Series D.

• Concentrated

  • Series A: 11.6%

  • Series D: 4.4%

• Broad

  • Series A: 5.8%

  • Series D: 2.2%

Now on paper, if you’re looking at a $1B company, then owning 4.4% is better than owning 2.2%. How sure are you that it will be $1B?

Marginal Confidence

Powerlaws dominate venture returns, where a small number of investments produce most of the earnings of a portfolio. This doesn’t just mean that only a small number of companies win. It’s worse: only a small number of companies a VC decides to back will return more than all the rest combined.

If you ask a VC to rank companies in their portfolio when they make their first investment, they will not do a good job. Consider their follow on funding. Wouldn’t it be better to just pick the head of the power law at Seed and make your extra investment then at a better price?

So now consider the ordering of those 25 investments. How sure are you that the head of the power law isn’t the 26th where you passed? Maybe if you made 50 investments, the top company in your portfolio won’t be worth $1B but $10B or even $100B. A 10X or 100X outcome is a bigger difference than only owning half as much at exit from lower concentration.

Does this mean that VCs are just monkeys throwing darts? Yes, but that’s the plan. Great investors show that their outcomes on average are consistently good despite not knowing whether an individual investment will matter.

I was reminded of this recently when folks on Twitter were reacting negatively to a prediction Clubhouse would be worth $100B in 10 years, when in fact it lost its footing after explosive growth. It’s normal to be overconfident and wrong, and just right on average.1

So the rules appear to be 1) more concentration means more ownership, but 2) on the margin you’re probably missing better investments.

In a way, this second rule proves too much: why not just keep backing more and more companies?

Well, Y Combinator actually settled on just that dynamic. They have fixed investment amounts, but a disproportionately large ownership at seed. They back as many companies as seem potentially great, rather than a target portfolio size. They scale the organization to match the company supply. I wish they let YC alumni like me be an LP because the model is awesome and effective.

Second Order Dynamics

There are other considerations over the long term. One I’ve noticed is that the founders you back become an asset to find new great companies. That matters intensely because within their domains founders often know a lot more than VCs. That’s why many transition from founder to investor. But the fresh network of founders even while they are running their startups proves to be an excellent way to find new great companies.

Portfolio size can also be a burden. How close tabs are you keeping on each company? How responsive can you be if you’ve backed two or three times as many founders? Does that hurt a founder’s interest in your fund?

If you’re spending management fees on helping the portfolio, then those resources could spread thin. If those tools scale with insider usage, maybe they have a network effect by themselves. For example, if you build a way for founders to get introductions through others in the portfolio, more is better.

I hope you can see why this issue is still actively debated. Some investors in funds will look at a basic regression from older funds and decide what works, but I think that doesn’t match the complex dynamics.

The Cruise story starts in grad school, where I focused on robotics, and then I worked on the DARPA Urban Challenge at iRobot in 2007. Around then I emailed Kyle Vogt of Justin.tv the first time because I was working on video streaming for robotics. They helped inspire my founding my first startup that got into YC like them. After that, I didn’t work in robotics for a while, so my ideas were building up. After sharing them over lunch with Ayden Senkut of Felicis, he said I should talk to Kyle, who had just started Cruise. We met, I loved the plan, and asked to invest in late 2013, pre YC. They sold to GM for $1B at lightning speed and returned 62X.

It’s funny when your first investment is that good. Investing is easy! Regression to the mean was inevitable. I founded Tango.vc in 2020 to back AI, ML, and Robotics companies. There were some challenges exiting ZIRP (higher rates mean less cash for late stage VC). Tango.vc Fund I has backed a few standout companies including Stoke Space and Freshpaint.

The Lovable story starts with the previous startup too. I met the team at Depict.ai in the S20 YC batch at Demo Day. Depict makes AI tools for ecommerce, like advanced search and suggested product bundles, and I loved their GTM tactics. The CEO told me a cofounder left because “he was AGI pilled.” That immediately piqued my interest to look into what he was working on. It was Anton Osika, working on GPT-Engineer, which looked incredible and eventually became Lovable.

I invested in the preseed in 2023. They started 2025 with a few million ARR, but scaled to $200M ARR this year. It’s such absurdly fast growth that it makes me double check the numbers. The product is incredible: Lovable lets non-technical people create internal tools, prototypes, and websites by just writing plain English. They’ve also just baked Lovable into ChatGPT, so you can build real apps right from their chat.

Lovable just announced their Series B. We invested out of Tango.vc Fund II and now this investment is the best in the portfolio. Our original investment is now worth 160X. The Net TVPI for Tango.vc Fund II is 3.7X. That means net of fees the Total Value of the fund is 3.7X the Paid In capital.

How good is that? Carta is a fund admin that tracks metrics that help compare performance. Usually funds take time to build value, so younger funds have lower multiples. I took their data about net TVPI across vintages and added Tango.vc Fund II which is well above the 90th percentile. It’s literally off the chart, which I enjoyed graphing below.

I’m currently raising Tango.vc Fund III. If you’ve like to become an LP, email me: ivan@tango.vc

And a heartfelt thank you to Anton and the whole team at Lovable. My job is really easy compared to yours, and I’m amazed by the pace of your progress.

For me, this whole story is so very Silicon Valley. I shared ideas with Kyle because the work at Justin.tv was interesting to me. Live streaming is not obviously related to robotics, but those that know the tech know the connection. Seed stage investing is very collaborative; Aydin didn’t need to share the insider knowledge about a new company. And then I met Depict at YC Demo Day. Their ideas and traction were excellent. But the best founders are compelled to build, like Anton starting the epic next company. The lesson for those aspiring: cultivate a sincere desire to build great things and always support each other along the way. That’s what makes Silicon Valley unique.

Subscribe now

Robots stole manufacturing jobs. Autonomous vehicles destroy trucker jobs. LLMs will bring mass unemployment. These are all catastrophizing and incorrect, and that’s before the Terminator tropes.

In reality, autonomy is a spectrum.

Understanding this matters for startups because the value of adding autonomy will move on a sliding scale, affecting how customers value your product and how you can price it. Product development also improves incrementally, where updated robots do more, increasing value.

Think of a scale from zero automation to full automation, with your product in between, moving right. Quantitatively, the ratio of human input value to total output value decreases, where more autonomy decreases the numerator and hopefully increases the denominator.

For example, floor vacuum cleaners were the first to get scale with robotics with iRobot’s Roomba. But they handle only part of the cleaning, so you still need human input. The first Roomba’s couldn’t self dock and couldn’t stick to one room. Those were solved. You still needed to empty the bin, and not until 16 years after launch did they add that. Why don’t you see Roomba’s in hotels? Because cleaning involves hundreds of different tasks, meaning we have a lot to go on the spectrum.

Teleo makes tele-operated systems for heavy construction equipment. Imagine a front end loader that needs to scoop from a wall, leveling a construction site, moving the load to be removed. The scooping and dumping are more dangerous and complex operations than driving back and forth from where you scoop to dump. It’s especially easy to drive if a human operator does it the first time and then the robot can repeat that same path. Then a single remote operator can manage two loaders at least, one that waits for input for scooping and dumping and the other that drives back and forth autonomously. An initial deployment will probably only involve a few tele-operated before scaling to cover one site and all sites. Even with that automated, who gets the vehicle to the site and performs maintenance? There is a smooth scale from all human to some human to no human for this job.

Amazon employs over a million robots to help with warehousing and shipping. They have robots that can move a stack of bins in a warehouse. But picking up something out of a bin and putting it in a box is harder than autonomously driving around an engineered environment like a warehouse. Autonomy scales from all humans, to some robots with automated storage and retrieval, to eventually robot arms packing everything into boxes. This is overly simplified, as there are dozens of tasks done by humans that haven’t yet been automated, like creating the warehouse itself.

Even driverless robot taxis like Waymo are incremental. First you have the scenarios where they’re allowed to be deployed, like at night, on freeways, or in bad weather (the Operational Design Domain, ODD). But even the marketplace dynamics show the spectrum because traffic is so famously spiky, with rush hours and midnight lulls. Unless we’re under heavy competition and ZIRP, robotaxis won’t scale to handle peak demand. So the spectrum moves from human only rideshare, to some autonomy, to more scenarios, and in the long run all cars can be autonomous.

And then consider robotaxi strategic finance. What if you wanted to cover all the peaks. That would mean average utilization goes down. The whole point of autonomy is to do more with less, but when you lower utilization, you lower revenue per robot. That means the payback period is longer and scaling out is more expensive. This is also why low interest rates with ZIRP would matter, because scaling is cheaper if capital costs are lower. This is why you might expect individual owners of robotaxis to meet the peak demand, not a centralized fleet. Individuals get other economic benefits from car ownership.

For a fundraising pitch for a robotics startup, part of the story is how long until you have pilot and scaled deployments. How many years before you have 1, 10, or 1000 robots in the wild? Add the autonomy spectrum to your plans. Then you can articulate how you’d spend more VC dollars on unlocking more of the market, and how your margins from increased customer value change over time. The best pitches don’t just describe a fully automated world, but the path you’ll take to get there.

We’re all accustomed to blustering arrogance. It’s the used car salesman overpromising the wares. It’s the sports fan whose team can do no wrong. It’s the professor who has spent decades in certainty.

Because startups are so hard, founders must have a firm self belief. So what kind of arrogance might that manifest?

The key difference is that you need to close. You need to have a product that works. Your predictions about the future need to be technically accurate and eventually realized true. This is a different kind of arrogance because unlike a lemon of a car, a Draymond Green flagrant foul, or the impractical academic idea, you need to be right.

I see this personality all the time in tech. And actually I love it, or at least tolerate it.

The challenge is that mimicry means you have a lot of wrong cargo-cult assholes. They think yelling and wearing a turtleneck is all you need to be Steve Jobs. They confidently take the credit of their engineering teams. They promise a feature on a sales call they don’t know if the team can ship.

At Tango, I need to judge the difference by closely listening to what the founders say. Conversation with founders is the primary way to judge an early stage company. Does your technology make sense? Will the market demand it? Can you find the right talent? Can you convince future investors? All require critical thinking about what founders plan.

Those confused by this might react negatively to the right kind of arrogance. These are the types of folks that dismiss Elon Musk as a know-nothing, despite the smartest folks in each of the fields where he operates disagreeing, and his dominance through Tesla EV sales and SpaceX mass to orbit. It’s arrogant, but it works. Everyone is fallible, but the quality shines through.

You can learn to adopt the right kind of arrogance by doing the work. It’s incredibly difficult to mime the confidence that comes from talking to hundreds of customers. If you’ve been heads down building, and you know the tech inside and out, then your competence will shine.

One failure case to avoid is hedging and caveats, which I most often see in academics. Research is so concerned with the narrowest specialization that academics will caveat and justify too much. Your company and potential investors aren’t a thesis defense, and you should just get to the point.

• I have deep technical experience with NYU, CMU, iRobot, Facebook, Dropbox, Lyft, and multiple startups

• We invest in preseed and seed stage startups in machine learning, AI, and robotics.

• We’ve backed over 70 companies. Our best investments include Cruise Automation, Stoke Space, Albedo, Freshpaint, Clone Robotics, and Lovable.

• We support founders with experience in engineering, product, growth, hiring, and fundraising.

• We write $100K to $250K checks

• We’re raising and deploying Fund III targeting $25M.

• We’re spending management fees to build software to help founders raise called Attention AI

• Get in touch at ivan@tango.vc

My Background

Venture funds are about their people, and Tango is a solo GP by me, Ivan Kirigin. So here is a bit about my background and experience.

I’m an engineer by training, with a background in robotics and computer vision from NYU and CMU. I worked on tracking, surveillance, bomb disposal robots, and self driving at iRobot for the Darpa Urban Grand Challenge.

This was before deep learning started working in vision, and because I wanted to ship products I did a startup outside robotics with Tipjoy in YC W08, one of the earliest cohorts. The YC founders and partners are great and I love that community. I then worked at Facebook where I understood web scale, and then Dropbox, where I helped early growth from 4M to 100M users.

My second startup was a mix of ML and social graphs with YesGraph in YC W15. Lyft acquired the company and I helped drive growth by building an ML suite to automate a $250M marketing budget. Automating a job felt like a finish, so I also worked in Lyft’s self driving division, Level 5. It was fun to see what changed (perception!) and what didn’t, like 3D scene structure from motion.

My first angel investment was in Cruise Automation in 2013, which sold for over $1B in 2016. It’s wild growth and exit so fast, but it figures that the first company to compel me to invest was very high quality.

I left Lyft to invest full time in late 2019. After Cruise, how hard could it be, ha! After some angel investments, I founded Tango.vc in 2020.

Fund Structure

Both Fund I and Fund II were around $5M, which is very small by VC standards. Now I’m raising Fund III targeting $25M – email me at ivan@tango.vc if you’d like to become an LP.

Fund III is actually already active, writing $100K checks into preseed and seed stage AI, ML, and robotics companies. This is unique, where most funds wait for fully raising before deploying. It’s a quirk of our LP history, and we’ll start raising Fund IV the moment we’re done with this marketing period for Fund III.

I started without keeping reserves in Fund I. That doesn’t work because it requires LPs to be brave when markets are afraid. Fund II kept reserves and also made initial fewer investments (24 vs over 50).

VC Math

The best way to understand VC is to understand powerlaws. If you look at a portfolio of investments, a minority will make up the vast majority of returns. One might be half of all returns, and the next two might be more than half of the remaining. An investment must be in this top set to make a meaningful return.

It’s worth breaking down the math to really understand the economics.

• Market Comparison: S&P real returns at ~8%, over 10 years, is 1.08^10 = 2.2X.

• Compare to median VC returns: 15%, 1.15^10 = 4X (Funds are illiquid so should return a premium over public markets)

• Top return: 4X fund * 50% of returns = 2X from top single investment

• Management Fees: 20%

• Investable: 100% - fees = 80%

• Initial checks : Reserves :: 70% : 30%

The remaining math could be just abstract and based on fund size and return ratios, but it’s easier to understand with typical numbers. I’ll use our model for Fund III.

• Fund Size: $25M

• First checks: $25M * 80% investable * 70% first checks = $14M. With 56 checks = $250K

• Follow on: $25M * 80% investable * 30% follow checks = 6M. With 24 checks = $250K

• Ownership: First check $250K / $20M post Seed = 1.25%, Follow on $250K / $35M post Series A = 0.7%

• Total Exit Ownership after dilution: 0.6%

• Target Exit Value: 2X * $25M / 0.6% ownership = $8.3B

• Multiple from seed to exit: $8.3B / $20M = 416X.

• 100X return is good. 1000X return is great.

If you’re interested in more detailed fund modeling because you’re an LP, get in touch for the full model: ivan@tango.vc

If you’re a founder who has complained about the home run bias of VCs, consider what assumptions you have, what specific things you’d change about this model, to make the math work. VCs that don’t believe in the powerlaw end up with bad returns because their top investments aren’t as large.

Our $100K investment in Lovable’s preseed is now worth the Fund II fund size of $5.3M, highlighting the real powerlaw. This deserves its own post, so subscribe to get it.

Subscribe now

What We Look For

In brief, the technology, the market, the go to market, and the team.

The best way I know how to assess these is to ask questions until I understand each. This questioning also measures intelligence, grit, and communication. My process is a lot more like an angel than a venture firm with an investment committee. I can’t imagine a worse structure to find outliers than a committee.

Technology. It’s a double edge sword, with harder tech serving a barrier to both you and competition. I try to understand this deeply, and my experience in AI, ML, and robotics means I’m often going deeper than less technical investors. I’ve worked to build many products, so how product strategy connects here matters. For example, you build an ML system, where in the product experience do you get feedback to refine and improve?

Market. Large markets are wonderful, but growing markets that will be large can be even better. It’s great to have a customer understanding to know how this might play out. Founders will find it very hard to mime the experience of having talked to many customers to know what they need. Specifically with robotics, I know and believe the default pitch, which helps to understand diffs.

GTM. The best seed stage founders have both modeled what it takes to grow while also having direct, measurable experience in execution. Traditional channels work all the time, but they require understanding of the execution required. For example, a sales driven company is almost certainly founder driven to start. There are some tells of inexperience, like modeling decreasing CAC with scale, which is almost never true.

Team. Founders building technology companies should be technical. Being technical doesn’t necessarily mean a degree from a university, but in the raw intelligence and grit required to ship something. Communication is an essential component here. You can actually mime this: write and edit down your messaging and then repeat the simplest version until you’re sick of it. Practice on customers, investors, press, your team, and potential hires. Confidence is often mistaken for competence, but clarity of communication from repetition and refinement is competence.

Supporting Startups

Founder asks are often interrupt driven, with investors called in to help. It’s useful to comment on routine investor updates too.

Technical. Founders are much deeper in the weeds of building their products than I can hope to be. But because I see a large portfolio, there are broad engineering and product lessons that could help. I’m also building a portfolio founder community where those that know best can help.

Hiring. This is usually process focused. I built a product in referral recruiting and worked to refine hiring wherever I worked. For example, I sometimes ask “did you tap your network on LinkedIn?” and founders will say “yes” and then I ask “did you export your whole team’s contacts and look at literally everyone for hiring and asking for referrals?” and the answer is never “yes”.

GTM. I’ve worked on multiple growth teams, so I know what excellent structure and performance looks like. The challenge early on is that you don’t have a team of dozens, so the founders need to trade action and attention to get the most impact. Modeling cross challenge cannibalization from a referral program using a multi touch attribution model is hard. Maybe just add a share link and a lifecycle email in an hour. Sometimes you just need to ask what is working and do that a lot more.

Fundraising. Introductions are common in seed rounds where people co-invest, vs later rounds where a lead will dominate the financing. I’m actually building a product to help founders raise called Attention AI. We automatically build your network, have a pre-built comprehensive investor list, and then help operationally get introductions. Communication around fundraising also matters, and I can help founders translate an often analytical and technical focused pitch into something less technical VCs can understand.

If you’re a founder building in AI, ML, or robotics – or if you are an LP who wants to back them, I’d love to hear from you: ivan@tango.vc

I was frustrated by some bad ideas, unanswered good ideas, and issues that weren’t even brought up. Here I want to explain how we should think about Truth and AI, and which of the ideas shared were worth preserving. A quick summary of the debate follows, then my commentary. They started with opening statements.

The Debate

Founder of Perplexity Aravind Srinivas, on the yay side, opened with the idea that because humans are innately curious, the tools will be used to feed that curiosity. Fitting for a search tool founder.

Journalist and author Nicholas Carr, on the nay side, said that AI summaries remove the need to think. Thinking is how we find truth, and we’ll end up with passable mediocrity.

AI Researcher Fei Fei Li, on the yay side, gave examples of answering questions for children and building smart systems to prevent the elderly from falling. She is a roboticist, so this embodied intelligence is a win, but not germane to a debate about truth. She teaches top talent and sees how they successfully use AI.

Technologist and author Jaron Lanier, on the nay, argued that our culture and business are setup for failure here. Just like social media isn’t inherently bad, the culture around social media makes us angry and anxious by using it. And on the business front, the ad business model means services push to sway us, and bend the truth. He also implored appreciation for complex systems with unknown side effects.

Fei Fei then argued that Nicholas was wrong because AI empowered the most thoughtful people. She also said AI isn’t the problem, but people, which was nonsense like arguing nuclear weapons aren’t a concern because people launch them.

Aravind pointed out that Perplexity is a subscription service at scale, so the ad business risk of swaying attention was lower. Fei Fei highlighted that competition means we’ll have higher quality over time.

Fei Fei also argued that the problem with generating an essay with AI isn’t that the student doesn’t need to think but that the assignment needs to be updated to improve how we assess student knowledge. That motivation is the key concern.

Bari Weiss, the moderator, brought in quotes like one from Marc Andreessen, about how AI will be 1000x times worse than centralized social media censorship because it will be one closed system. Aravind said open source will help, while Jaron correctly pointed out that even open weight models are inscrutable today. Just today Grok open sourced their system prompt after an employee injected disinformation.

Nicholas concluded with a plea for artists. They are in touch with truth, and their work will be worse when AI synthesis is plausible, generic, and soulless. Jaron echoed this with a concern about attributing creator work in the models, and getting them paid fairly.

Aravind said early that he didn’t know how to debate and that AI helped him prepare. He ended the evening by reading a prepared conclusion that said “the other side wants to paint a dystopian picture.” Nicholas’s concern about passable mediocrity was thus proven correct. Instead of listening and addressing any issues live, Aravind read an AI-prepared plausible conclusion against a phantom.

The yays lost the debate, measured by how much the audience swung from yay to nay. The evening started with 68-32 yay Truth will survive, and ended with 45-55.

Going Deeper

I’m firmly in the yay camp, so let me make some arguments that Fei Fei and Aravind didn’t.

Truth is not a state of being but a process in our society. We value truth seeking, and many cultural institutions embody that value. To answer the question, we should look at these institutions in turn:

• Education, how we train and assess the next generation

• Research, the pinnacle of formal education, where we find new knowledge

• Engineering, with technology, how we deploy and scale our knowledge

• Security, how we protect what we value

• Journalism, with media, how we explain what is going on

• Art, as emotional and poetic truth

Education

The default view from educators is that a 5 paragraph essay can now be produced by AI and thus AI is cheating. Take a photo of a math problem, and get the answer. Use a browser extension to do both, and let the AI take your test. Each is very bad, obviously. Kids won’t learn the content or how to think. But like Fei Fei mentioned, the assessments are at fault. In education, the Bloom 2-sigma effect notes that 1:1 instruction is two standard deviations better for education performance. That isn’t affordable, but with AI we can achieve it. Our education system will adapt to deliver the best educational outcomes. Some institutions might resist and slow this change, like teachers unions or school district managers, both which resist any changes. We spend close to $20K per student per year, but a $2K/yr and maybe a $200/yr AI might outperform our teachers. AI that can understand your level of knowledge, tune explanations, and have infinite patience just for you.

Fei Fei was correct about motivation being key. It’s the intrinsic motivation to improve that will compel kids and parents to choose the best AI tools. Unfortunately this is a hard topic because do we even know what motivates or demotivates kids? Some think we need student driven inquiry. We also need heroes, and in our age of miracles we’re getting more and more examples. Getting humans back on the moon and to Mars. Colossal bringing back dire wolves. Boyan Slat building machines to clean our oceans and rivers. Can a personal AI tutor deliver personal motivation to achieve the impossible?

That is hard to predict, but I know already there is a Cambrian Explosion of tools custom to task in education.

Research and Engineering

What’s very clear already is that AI has an unambiguously positive impact on research and engineering. This wasn’t brought up in the debate much at all. Software engineering has gone through a revolution where regular people can ship whole apps, like using Lovable. And trained software engineers can produce and deploy code 10 times faster, with Copilot, Cursor, and Codex. AlphaFold has released 200M proteins, which represents a 1M times speedup in biotech research.

Security

Imagine your grandparents get a video call with your voice and face saying you got in an accident while traveling out of the country and they need to wire money. Deep fakes like this are a major concern. This predates AI, with cybersecurity broadly: from this social engineering to breaking into systems and ransoming data. Security is a cat and mouse game, and it’s not obvious AI offense will beat AI defense. Your bank should have systems that scan for weird wires. Maybe even your communication tools should have AI authenticators that detect fakes. Maybe signing all messaging is required to verify yourself what is said.

Journalism

Before printing, people lived in a media vacuum. From the very start of printing, bias was part of the game, from the American pamphleteers to yellow journalism. It was only with monopolies from high powered radio, huge newspaper printing machines, and restricted television broadcast channels that the news business boomed to the extent that editorial staff could afford objectivity. Then the internet exploded this model and we’re still trying to figure out what it means. The best media sources today have declared editorial perspectives while maintaining truth seeking. Unfortunately, bad faith and outrage on social media mean many are successful voices that spout nonsense. Concerns around deep fakes are overblown because skepticism and tribalism dominates our media today. We already believe fake things that don’t use generated imagery.

What few are talking about: personal AI to direct your attention. Maybe people are thoughtless, but our distracted minds are downstream from centralized algorithmic feeds optimized for your engagement and platform ad revenue. While I care about businesses efficiently reaching customers, I don’t need my feed influenced by anything other than what I want. AI is finally good enough to take a high level description of your desires, and to feed you that. This is on the consuming end and can’t be blocked by platforms. If you stream it, you can filter it. System 1 thinking is fast and intuitive while System 2 thinking is slow and deliberate. Algo feeds and push notifications thrive off System 1. But Personal AI feeds will be driven by System 2 intentional desires.

Concern around social media is also a current moral panic driven by narratives that lack real data. We talk about rage inducing tweets. What about the millions of hours of content that feed your soul. Rick Beato music theory. 3Blue1Brown math wonders. Nerdwriter screenwriting. Crash Course history. Mark Rober’s inspired engineering. Here Jaron is right that complexity rules. Media is so large with so many niches that it’s very hard to argue we have a good enough grasp to judge. Humility should defeat a moral panic. We should be most concerned with what generates the thoughtfulness to seek out the gems. We need to fight brain rot, AI generated or not.

Art

Generating images, music, stories, and movies is getting easier and easier. This is a double edged sword, with content creators potentially losing livelihoods while more and more people can finally create. The struggle to master a craft might be inherent. Learning music theory is actually just hard, no matter our best efforts to make it easier. But to edit video in tools like Adobe Premiere takes many times the length of the source material, and that struggle is just a deadweight loss. Sometimes better tools just help us achieve our vision better and faster. We also have business conditions that warp what artists receive. For example, very few people make a good living off streaming music from Spotify. But everyone in tech will tell you that is downstream from the behavior of music labels, not inherent to the technology itself. Music has the most extreme ratio of emotional salience to monetization, the inversion of sports video games. And then you have trends like Marvel or Star Wars mediocrity, where studios spend every resource possible to generate dreck.

Even without AI, content volume has exploded, and that will accelerate. So the question is on meaning and monetization. I predict more people will find personal meaning in their own creativity and that of others than meaning lost from a change in the caliber of content overall. I’d also predict more people make a living off of creative enterprises than do today.

We raised Tango.vc Fund I starting in 2020 and that fund is now fully deployed. There are over 52 investments across 46 companies. Fund II is closed, and we’re in the middle of deploying it. We’re writing $100K to $250K checks in preseed and seed stage startups. 

Read on to learn about some Tango.vc Fund II companies, 3 lessons learned, a note on Fund III, and news about my new startup.

Fund II Companies

Let’s walk through a few companies currently in Tango Fund II. This isn’t a complete list, but gives you a sense of what we like.

Some companies are follow up investments from Fund I, like Vici Robotics which makes shelf stocking robots and Clone Robotics which makes humanoids. Check out Clone’s grip!

Some companies are new, but with founders from Fund I. The insight gained from working directly with a founder is impossible to replicate. Lovable is making software development AI. PitPro is building robotic fleet vehicle maintenance.

Ion Design can turn a Figma design into working React code. I’m personally eager to level up with their design system too.

Adagy Robotics helps companies with autonomous robotics save the robots with remote teleoperation when they get stuck. 

Superfocus is building AI that doesn’t hallucinate, which is crucial in cases like customer support. 

Terraform Industries makes methane from direct air captured CO2, all solar powered. This means methane, an important industrial input, can be climate neutral.

1. Reserves Are About Conviction

I had planned to invest in later rounds via special purpose vehicles (SPVs). My thinking was that building up a network of founders across more companies was the best use of funds. It also seemed weird to have multiple stages of investing, starting at Seed and following up at Series A or Series B. 

Unfortunately the pull back in the market was timed with some great opportunities. With SPVs, the LPs decide per deal which investments to do. Right when I had the most conviction, the LPs would be the most conservative. 

This means having reserves to follow-on is actually much more about conviction than anything else. The overhead of convincing LPs per deal is fine when times are frothy. This LP strategy can best be described as “buy high”, which is unfortunate. It’s better for LPs to trust the conviction of fund managers and dedicate some of the investing to later rounds.

2. This Is So. Much. Fun.

Talking with founders daily is such a delight. I can do this daily, for the rest of my career. This is true even when there are challenges, where helping solve problems is meaningful. It’s especially fun to see the very earliest stages progress onto real traction and incredible products. 

Sometimes it’s visceral, like watching Stoke Space run a hop test for a whole new kind of rocket. Go see the video here. Currently there are 3 companies that have ever flown reusable second stage hardware: NASA’s Space Shuttle, SpaceX’s Starship, and Stoke Space’s hopper test vehicle. Go watch the deep dive on Stoke here after watching the hop below.

Other times, the drama is from the metrics, with some SaaS companies working incrementally to then find explosive growth. The hockey stick is great, but seeing it after years of heads down product and market development hits differently.

3. Know Your Tech

It’s wonderful that ChatGPT is so obviously good that many are working to build in the AI space. This appears to come at the cost of founders choosing robotics and hardware companies. YC batches went from dozens to single digit robotics efforts. But the opportunities in pure software are real and compelling. 

With a buzzword-filled trend, you need to watch out for people that don’t know the tech. It’s now more important than ever to be able to understand what is possible and what isn’t. And all the startup fundamentals still matter.

For example, some investors dismiss products by saying they’re “ChatGPT wrappers”. This is naive because you can fine-tune open source models to beat OpenAI on price and performance, and then you’re still using AI like an API call either way. But this also ignores so much else that matters, like access to proprietary data from customers, the problem in the market you’re solving, the quality of the user experience, how you reach customers, and so much more. It’s like dismissing a company because they use MySQL. 

Fund III Coming Soon

We’re no longer raising for Fund II, but my plan is to raise Fund III starting at the end of the year. Like Fund I, this will be a 506c fund, which means I can openly talk about the fundraise. 

The main reason to make it open is that the network of LPs is valuable for companies, so getting folks from Twitter and Substack makes for surprising new friends. 

You can expect the same focus on seed stage AI, ML, and robotics. 

If you’re interested in investing in Fund III, please fill out this form.

One More Thing: My New Startup

In addition to running the fund, I’m now running a new startup, Attention AI. This is my third startup, and I’m excited to tell you more. But that’ll be in a future post. So don’t miss out and subscribe to learn more soon. 

Subscribe now

What follows is an introduction to venture economics and startup pitching. This is based on a presentation, so I recorded the talk here, which you can read below.

Subscribe now

Credentials

I studied computer science at NYU and robotics in grad school at CMU. I worked on bomb disposal robots at iRobot, before starting my first startup, backed by Y Combinator. I then worked in engineering at Facebook and product at Dropbox. My second startup was acquired by Lyft. My first angel investment was in Cruise Automation. I founded a new VC fund Tango.vc and just founded my third startup, Attention AI.

Types of Investors

Different types of people invest in startups.

Often founders will get friends and family to invest. They are low sophistication, investing their own money because they like you. The first concern is to make sure this unsophisticated buyer understands that startups are high variance and have a good chance to be worth nothing. The second concern relates to rules from the Securities and Exchange Commission, the SEC. In short, only take money from Accredited Investors, which essentially means having enough money to lose.  

Angels are sophisticated but still investing their own money. They should already understand startups and their risks. They are almost certainly already Accredited Investors. The key difference between VC and Angels is investing other people’s money or your own money. 

Venture Capitalists raise money from Limited Partners (LPs). They are limited because they have no say in day to day investing decisions. VCs have partnerships that decide how to invest. This means convincing a few people to invest. VCs will have a lot more money to invest than angels, so larger funding rounds focus there. 

To understand what convinces VCs to invest, you need to understand the economics of their business. That means understanding carry, illiquid returns, and most especially power laws.

VC Economics: Carry

VCs make money in two ways. For assets under management (AUM) VCs earn a small percentage in a yearly fee, typically 2%. This pays salaries and other costs because startups take so long to pay off. VCs return cash or stock to LPs once the startups they backed go public or are acquired - years after the initial investment.  VCs take 20% of the profits, called Carried Interest. “Carry” derives from shipping, where captains would take a profit share from goods they carried.

VC Economics: illiquid returns

If I lend you $100, how much should you expect back in a year? The US “Risk Free” rate for bonds backed by the US federal government and thousands of nuclear weapons is currently 3.88%. The bonds are liquid, as are other assets like public stocks, which means you can sell them at any time. The long term rate of return for the S&P 500 averages 7.38% after inflation. 

Investments in startups are illiquid for years. This means the rate of return should be higher. This percentage growth yearly is called Internal Rate of Return (IRR). Over 10 years, IRR averages compound into a large multiple: 8% = 1.08^10 = 2.16X. 9% is 2.36X, and 10% is 2.59X.

This means a whole fund must return multiples of the original to justify the illiquidity. 

Actually, many funds do much worse. The graph below shows the top and bottom quartile of fund IRR by the year the fund started, showing that some years the bottom funds return less than what was invested.

VC Economics: Powerlaw Returns

Each VC fund will make dozens of investments. But only a small share of those investments will impact the total returns. This isn’t intuitive because powerlaws don’t match a lot things in the world. One home on a block isn’t worth all the rest combined. One person in a crowd isn’t as tall as the rest combined. But for VC portfolios, it isn’t unusual for a single investment to be worth all the other investments out of the fund combined. 

Let’s illustrate with an example. Accel invested $12.7M for 11% of Facebook in 2005, valuing the company at $115M. In 2010, they sold 20% of the stake at a $35B valuation worth $500M. $500M / 20% = $2.5B ~= 200 * $12.7M, meaning the investment already returned 200 times. In 2012, Facebook IPOed at $90B earned another $4B for Accel, totaling 350X. This means the original $440M Accel Fund IX returned 10X just from Facebook. 10^1/10 = 1.26, which means an IRR of 26% from this single investment. From the graph of fund distributions before, this single investment made the firm a top performer. It literally doesn’t matter what happened to all their other investments. 

The fundamental conclusion here is that the very best investments a VC makes matter, but all the rest don’t. This means every single VC investment must have a chance to be the top. 

See this illustrated here, where the fund on the left hit a bigger head of the powerlaw, but the fund on the right had fewer zeros, with more companies returning 1X or 2X. Fewer losers might seem better, but the fund with a better head of the powerlaw is worth twice as much. 

To get VCs to invest, you need to convince them there is a chance your startup will be the best investment they ever make.

Startup Attributes

Because startups take years to take off, investors need to predict the distant future. The best evidence of future stellar performance is past performance, meaning traction to date. But what might predict more traction. Early on, you might just have an idea and the team. Here are the dimensions to consider. 

Team: Who are the founders and early team members? Where have they worked or gone to school? What have they built in the past? Do they seem relentlessly resourceful?

Technology: What technology is required? Is it unique to the company? What work is ahead or already proven?

Market: How big is the market? How big will it be? How fast is it growing now? Is the current shape fragmented, a large incumbent, or something else? 

Traction: Is your product available? How much money are you making? How many people are using it? Do they love it? 

Go To Market: How will you reach your customers? How much will it cost? How fast can you grow? Who is the buyer? Are there regulatory hurdles? 

Culture: What makes your company unique? How will you keep your culture strong? 

Competition: Are there other players? Large or small? What advantages or disadvantages do they pose? 

Why Now: What has changed that means your startup can succeed? Why have others failed before?

Bad Pitches

Here are some red flags to watch out for within these categories. If your startup suffers from them, try to solve the problem or at least have a good answer. 

Team

🚩 No one technical

🚩 Everyone part time

Technology

🚩 Tacked-on trendy buzzwords, like AI or blockchain

🚩 Outsourced development

🚩 Hammer in search of a nail, like a grad student that just finished a thesis, looking for an application of the technology.

Market

🚩 Contingent growth where your customer growth isn’t guaranteed, like battery management for satellites

🚩 Substantial regulatory barriers

Traction

🚩 Shy or uninformed about metrics

🚩 Cumulative metrics. Always show rates of change

🚩 Graph manipulation, like compressing an x-axis to seem faster.

Go to Market

🚩 "We spent $0 on marketing" probably means you don’t understand the cost of different channels. 

🚩 Modeling CAC decreasing substantially

🚩 Not metrics driven

🚩 Founders want to hire head of sales before establishing repeatable sales model. Symmetric with "no one technical"

Culture

🚩 Playing house

🚩 Not working long hours

🚩 Anything dishonest or low integrity

Competition

🚩 Inaccurately portraying your competition as hapless, unloved, and expensive.

There is a lot more to say about all this, but I hope this was a good introduction. In future posts, I’ll write in more depth about both VC economics and pitching. 

Subscribe to get future posts.

Subscribe now

Around the time of the American Revolution, Thomas Malthus predicted massive famine. In 1798 he said, “The power of population is so superior to the power in the earth to produce subsistence for man, that, unless arrested by the preventive check, premature death must in some shape or other visit the human race.”

The thing is, he was correct. Looking at contemporary data, you’ll find that population and wages are inversely related. You can make people richer only if there are fewer of them. There is a natural capacity, like we see elsewhere on earth.

This turned out to be wildly incorrect since the time of Malthus, when both population and wages have exploded. 

Why? Yes, the industrial revolution, but what specifically? 

First, you have a steam engine, powered by coal. One of the first applications was to pump water out of mines where you could get that coal. Then you could also get iron ore. Coke is made from coal, and when combined with iron ore, you get steel. Steel is used to make engines and rail lines and more. Sure, this is great for industry, but how did the population get supported? Transporting fertilizer by rail line is the specific way farms were able to produce enough food for the exploding population. 

Fast forward to the 1960s, and you had a new version of Malthus in Paul Ehrlich. In 1968 he said, “The battle to feed all of humanity is over. In the 1970's the world will undergo famines — hundreds of millions of people are going to starve to death in spite of any crash programs embarked upon now. Nothing can prevent a substantial increase in the world death rate.”

Nothing like this happened. Not only did the population increase, with a smaller share of people living in extreme poverty, but also the absolute number of people living in poverty went down. This is astounding progress.

Too many people know about Malthus and Ehrlich, and not enough know about Norman Borlaug. Borlaug and other industrial agronomists helped create and popularize high yield varieties of wheat and rice. At the same time, the best techniques became widely used, including chemical fertilizers, pesticides, herbicides, fungicides, mechanization, containerized global trade, and bulk shipping. We also went through a health revolution with lower infant mortality, better education, and access to contraception. 

Technology that lets us do more with less is the fundamental driver of progress. 

Technology drives progress. Building this directly through new startups works exceptionally well because new companies aren’t encumbered by practices in the past.

Venture capital is a lever on technology. The reward for doing well in VC is getting a longer lever. Success is rewarded with scale. This is my fundamental motivation in working in VC.

A Network Syndicate lets a high affinity network of highly experienced alumni from schools and startups invest in itself. 

1. Exclusively alumni LPs backing exclusively alumni founders 

2. Both a rolling fund for the index, SPVs for per deal

3. Algorithmic investing based on distributed LP decisions

4. No carry or management fees

Let’s back up: experienced operators have been investing as angels for a long time. They have direct experience and cash from their success. More recently, some have raised funds as leverage, and many successful CEOs have funds on the side. We also have seen some alumni run SPVs and AngelList Syndicates.

These platforms don’t exclusively have alumni LPs, don’t exclusively invest in alumni, and don’t invest in every alumni deal. They also charge carry, despite the initial value creation coming from the original company or school. Many alumni work at VC funds themselves, whose LPs don’t want to pay carry in both the alumni vehicle and the fund. This lowers the overall funding potential for alumni.

I’d like to see a Network Syndicate for my tightest networks: NYU, CMU, iRobot, YC, Facebook, Dropbox, Lyft, but I’d also hope to see this for other universities and companies. If you’re interested in getting involved, complete this survey: 

https://airtable.com/shrs4CGxuh8K1c1B3

Let’s define the pattern:

1. Exclusively alumni 

All the LPs are alumni and all the investments from the pool are in alumni. Contrast with large syndicates today, where founders need to accept money from a large pool of anonymous LPs. Unlike a VC fund, these LPs need to see their detailed pitch to decide to invest. Alumni increase affinity and lower this cost. Alumni are also more likely to help the founders they back. Backing non-alumni founders dilutes this network, with ample other ways to invest in such companies. You can still work at the company too!

2. Rolling Fund + Syndicate

I’ve talked to other folks that manage networks. Some people just want to support every company, without thinking too much about it. Others want to learn how to assess companies, and would only invest per deal. Many start with per deal and move to the index. 

We’ll start this experiment as an AngelList Syndicate.  If there is interest from LPs and even a single deal closes, it’s worth spinning up a Rolling Fund. Rolling Funds have a subscription which allows people to contribute over time, which matches how normal people manage savings.

3. Algorithmic Investing

There is some overhead to setting up an SPV in a syndicate. This means enough people need to back the syndicate in order to close. The wisdom of the crowd can decide on the deal. If enough alumni LPs want in, the deal happens. This pitching, sharing deal details, and vibrant group discussion, but the “algorithm” is literally a tally of commitments. 

4. No carry or management fees

A background on VC fees: VCs pay their salary with ~2% fee yearly on assets under management. Startups take a long time to exit, so you can’t count on profits from deals themselves. Those profits are the other source of revenue, where VCs take 20%, called carried interest. If someone is investing full time, finding deals, and supporting companies, then this makes sense. But if the decision is distributed among LPs and the LPs are there because of a common connection, neither fee makes sense. There are day to day operations to run these groups, but there is no shortage of people that want experience with investing that will help here. 

I personally run Tango.vc and would love to be more systematic about the deals I see with alumni. Better deal flow is all the compensation I need because I’ll get paid in Tango.vc carry.

If you’re interested in getting involved, fill out this form here. 

https://airtable.com/shrs4CGxuh8K1c1B3

My plan is to explore the CMU and Dropbox networks first because I see opportunity there. If a lot of folks are interested in other networks, I’ll help you get organized too. Share this post with friends who could join.

Share

One end of the spectrum is something like B2B SaaS. You’ll be judged by well-known performance metrics. The example below is from Initialized. They know so much about these types of businesses that they can give thresholds for metrics to earn a Series B.

Another end of the spectrum is very Hard Tech. Think artificial general intelligence (AGI) or launching rockets. Tesla was founded in July 2003 and shipped its first roadster in July 2006 - a full 3 years. SpaceX was founded in March 2002 and had its first successful commercial launch after 3 failures in September 2008 - 6.5 years later. Deepmind was founded in 2010 and acquired in 2014 without a public product beyond their incredible research.

It’s actually relatively easy to pitch these most ambitious hard tech startups. Venture Capital bets on outliers, and the story “if this works, everything changes” fits very well. It doesn’t have to be a large chance of winning. A VC is happier with a grand slam home run than getting on base many times. 

The lonely middle on this spectrum is different. Your company has deep technical challenges that might take some time to solve. But it might not be judged like solving AGI. Instead, you might be judged compared to B2B SaaS metrics. 

Imagine year 2 of SpaceX or Tesla. New Revenue: $0. Growth 0%. Retention: who knows?

At Tango, we back a lot of companies in the middle here. Traction can solve problems, but short of that without a launched product, you have to find proxies for the value you’re building. 

Early interest can take a few forms. Letters of Intent (LOIs) are non-binding contracts that share a customer’s interest in purchasing. Because they’re non-binding and in the distant future, the numbers can get quite large, especially in large markets. I’ve seen “hundreds of millions in LOIs”. If you map the contract value per customer to the whole market, you get a large addressable market. 

If you’re close enough to launch, selling first units before they are ready can pull forward traction. This requires the customer interest to be strong enough to put down money, maybe only a fraction of the purchase price, early on. The cash is nice, but the signal is crucial. 

Sometimes customers can pay you to develop the product they want to buy. It’s normal for larger companies with enterprise clients to have service revenue as part of a deal – maybe as high as 30%. Consider adding this revenue before you have a product to show that customer interest is there. Take care in how you talk about this revenue to investors because it isn’t ARR. Nothing spooks investors faster than telling half truths about your numbers. Investors also don’t like consulting and service businesses, so you need to make this message especially clear about how this service revenue is helping hit technology and product goals. 

One of the clearest ways of demonstrating you know how to break into a market is by showing how you understand the customer. This comes across in many ways. How many have you talked to? Do you have a causal explanation for why they want to solve a problem with your product? Can you quote them on it? 

Customer understanding is especially important in industries where tech has lagged. It’s precisely this lag that creates enormous opportunities, but also investors have heard the story too many times. This includes industries like construction, education, energy, and agriculture. It’s not enough to say that “construction productivity growth is actually negative”. You need to explain why your product will change that. 

Subscribe now

A few Tango.vc portfolio companies come to mind that live in this lonely middle of the hard tech spectrum.

Teleo builds autonomous and teleoperated heavy equipment. Think about moving large amounts of dirt and rocks on construction sites or mining operations. The economic case is crystal clear: operators are aging and expensive and allowing them to spend more time piloting is a win. The customer story needs to spell out much more: how does rollout work? Do you have a support network for the equipment? How are operators trained? Answering this and more helped earn a Series A for Teleo. 

Faction makes driverless deliveries. Driverless vehicles are famously difficult, and some would argue on the far end of the hard tech spectrum. But Faction is built different. I’ve never met a robotics founder more obsessed with making a profitable business and launching a real product than Ain McKendrick from Faction. Every design choice feeds this goal, including choosing a light electric vehicle chassis, avoiding human passengers, and leveraging teleoperation. Their answer is to have real revenue from real deliveries as soon as possible. 

Edia makes AI software for math education. The product helps generate math questions, can grade them automatically, and give explanations for where students went wrong. There is a deeper layer to Edia, evident from Joe Philleo’s twitter bio: “solving Bloom’s 2-Sigma problem for K-12”. The Bloom 2-Sigma problem is the notable impact 1:1 education has on performance. 2-sigma refers to two standard deviations of improvement, which in the world of education is staggering. But humans are too expensive to do this: in the US the ratio of kids 5-18 to working adults is 33%. Software can solve this problem in the long term with AIs that are as good as real teachers.  

This is an enormous ambition, but what do you sell right now? I’ve learned so much from Joe about the education market, specifically in public schools. What will parents pay for? Who makes decisions not for schools but for districts? What is a district’s “IT budget” per student, and how might that thinking evolve? What measure of student success from a product turns into district enterprise contracts? The level of customer understanding here is crucial to build and sell a product well before you have an AI tutor for every student. 

In summary:

• Keep your big vision. Investors love outlier hits.

• Build revenue signals like LOIs & pre-orders

• Consider services revenue, but keep it connected to product roadmap

• Build customer connections and your understanding of their industry

• Clearly model why a customer will buy, what problem it solves

• GTM includes where to buy and where to get support

• Move as fast as you can to early traction. Traction solves all problems

I’d love to hear more tactics and strategies that work for other companies on the hard tech spectrum. Add them in the comments or email me: ivan@tango.vc

Subscribe now

It’s been a journey to get here. I started investing full time 2 years ago, and I wasn’t sure if I was going to run a syndicate, stay an angel, or raise a fund. As I learned more, it became clear that there are way too many opportunities for a new fund to do things differently.

Fund I is all about proving the basics: raise money, find great companies, win the deals, and support the founders. My angel investing before the fund formed is now rolled in, so we’ve actually already backed over 25 companies, with portfolio value at 1.7X. This is wonderful performance, especially considering how much hard tech we’re backing. 

Today we’re also announcing the Tango Community. If you’re reading this, you’re already a part and you should subscribe.

Subscribe now

We have some big plans for the community. Founders and LPs are already using the channel to connect and share deals. 

While Fund I is done, you can join our syndicate for follow-on investments. Tango will back the seed, and the syndicate will back Series A and beyond. 

Tango Fund LPs get discounted carry, and carry is paid back to Fund I. This carry flow is unique to Tango, and our modeling shows this carry could add a 1X multiple on Fund I. This is a good example of optimizing for the long term because the performance of Fund I and II will impact Fund III fundraising. I’d rather boost Fund III later than pocket the carry personally sooner. 

There is a lot more we want to do differently, from how fund carry flows in a partnership to how management fees should be spent on software and services. Subscribe to learn more in future posts.

Subscribe now

In March we started fundraising in public and the response has been stellar.

We added over 50 LPs and went from $1.6M committed to $3.8M. Regardless of the final amount raised, we can execute on our strategy, which feels great.

The most exciting part is that the community building from small LPs is going to be a superpower as we scale. Our friends over at Weekend Fund put it well: "1000 True LPs", a play on Kevin Kelly's 1000 True Fans. I've been a fan of this idea since my first startup, Tipjoy, which enabled social micropayments for creators.

But this can run into limits based on SEC rules about the number of investors. Weekend fund solved it by having a parallel fund. I'm going to solve it by keeping the minimum $25K for individuals while setting the minimum for institutions and family offices at $250K. I also plan to raise just under $10M, so the SEC limit on investors goes from 99 to 250.

For funds & family offices: $250K minimum https://angel.co/i/aPEzB

For individuals: $25K minimum https://angel.co/i/xXmYi

There is a lot more coming shortly, with things to announce soon around recruiting, and lot more content to help illuminate the world of startups and VC.

Subscribe now

I sat down with Andy to dive into the industry - from the post Apollo nadir to the rise of commercial contenders. Below is the video, followed by what I learned. 

If you like this and want to see more interviews like it, head over to YouTube to like, subscribe, comment and share. 

Subscribe now

Andy got started with a PhD in Aerospace Engineering, and then spent 10 years building rockets at Blue Origin. This culminated in building Apollo sized rockets, which is going to send very heavy loads to orbit and the moon. 

Large rockets face a tradeoff with rapid reusability. Going to orbit means flying not up but over, and returning the vehicle means taking more fuel. This broadly means there is a gap in rocket designs, and Stoke wants to build 100% reusable rockets designed to fly daily. Servicing the satellite market isn’t the same as getting humans to orbit and beyond. 

Launch costs used to dominate costs and building for space was very expensive. But now sensors and compute are cheaper, which makes the launch costs a higher share of total costs. This is pushing costs down, especially because private companies are meeting this shift in demand. 

Timing is interesting. Could we have created this market after the moon landing? It would be possible, but the biggest cost would be the army of technical folks doing what is now done with CAD, computational fluid dynamics, finite element analysis, etc.

The government that pushed Apollo still has a strong role. They should make space secure, both physical security and the law. The four pillars of growth are all moving to space: communication, transportation, energy production, and manufacturing. See this video with our interview with  Varda cofounder Delian Asparouhov. NASA is also beyond our scientific research and exploration and should keep that important role.

If you don’t have the high ground, defense is very difficult, including here on earth. It’s hard to imagine the US military without communication, GPS, and surveillance, but that’s what will happen if an adversary can take out satellites. Symmetrically, we can dominate authoritarian powers like China and Russia if we have those offense and defense capabilities. 

Venture Capital should understand that hardware takes longer to mature than software, but you need to back those companies because they interface with the world. That includes manufacturing, construction, logistics, energy, and farming. 

We touched on motivation and hiring too. Andy naturally was drawn towards building things, especially explosive things. More recently, this turned into an obsession with commercial space, which is a building block of sustainable multi-planetary life. Examples of visionaries help inspire the next generation to try to change the world. Taking a chance on new talent helps propel people into their careers. 

When hiring, it’s important to have clarity of vision in addition to technical skills. You need a mix of hunger and experience. When you have a solid foundation in your team, you have the support required to bring in people with less experience. Note that this applies for something as technically challenging as rocketry, but software scales to less experience better. 

Remote work for hardware is difficult because at some point you need someone on the ground. Having a whiteboard to work through a problem also seems essential. Beyond that, a team needs a tight coupling between design and testing. It’s actually the most important part of moving fast. 

You must test your designs before going so far in the wrong direction. There is actually a pejorative term in aerospace: “PowerPoint rockets” -- meaning designs that are very far from seeing the light of day. 

Starship is an amazing recent example of being hardware rich, where there is a fast turnaround to test after a failure. 

If you want to work at a company like Stoke, Andy has only one piece of advice: build. It’s important to get technical skills, but nothing will help more than trying to get something into the real world. 

Tango.vc backs great companies like Stoke. If you want to learn more about our VC fund, we’re actually fundraising openly.

Learn more about our fund: https://www.tango.vc/p/fundraise

Back Tango.vc: https://angel.co/i/aPEzB

Subscribe now

Let me know who you are to learn more. We held an open LP Q&A, and here are the notes.

This post covers:

• Why I can now do this in public

• Overview of the fund, including my experience, fund structure, what founders say

• How my 1st angel investment turned $25K into $1.5M

• What to expect in future posts

• Future posts going into detail about software, services, incubation, SPVs, partnerships, and more

This kind of announcement is rare because of the rules around fundraising. In this post, I want to explain what changed. I also want to outline our plans to help folks understand what we’re doing differently at Tango.vc. Our ideas don’t fit in a single blog post, so you can expect many more posts on this over the next few weeks. 

First, the rules: the SEC wants to prevent naive investors from losing money. They require firms to register (like the overhead of going public), but have exemptions. 506b mandates that you don’t advertise and all your investors are accredited. 506c allows for advertising if you verify that accreditation. Tango.vc just switched from 506b to 506c. This means that unlike most funds, we can write about the process more openly. 

You might have heard of AngelList’s rolling funds. Those are also 506c, but coupled with a “subscription” instead of a discrete vintage. Tango.vc Fund I is not a rolling fund, in part because I believe the fund mechanics make more sense in the traditional way. LPs that commit will just have their total commit amount to consider.

I’m so excited to share more about how VC works. Fundraising in public means much more content will be open on this blog.

To start, let me share what has been stuck in private fundraising conversations: my experience, track record, the fund structure, and more.

I’ve been obsessed with AI my whole life, starting with loving SciFi, from Data & Deckard to Hal & Mr. Smith. It’s why I studied computer science. I quickly learned that SciFi lies and nothing works yet, so I wanted to learn more and fix that.

At Carnegie Mellon University I got a Master in Robotics, focused on computer vision for field robotics. I worked at iRobot on the research team, which included helping the DARPA Urban Grand Challenge for autonomous vehicles. 

In 2007, I decided there wouldn’t be new robots for a while (which proved correct), and founded Tipjoy, a social micropayments startup. We were backed by Y Combinator in Winter 2008. We ultimately shut down, and while at Facebook in 2009 & 10, I learned what real growth looks like. 

I joined Dropbox to help them drive growth. I was #22 as one of their first product hires. Dropbox grew 12X in 2 years. I helped build data systems and run dozens of experiments around sharing, referrals, and onboarding.

In 2012, my 2nd startup YesGraph focused on machine learning over social graphs. We were backed by A16Z, Accel, YC, NextView, Bloomberg Beta. Lyft acquired YesGraph in 2017. At Lyft, I focused on user acquisition, building machine learning systems to manage $250M+ in marketing spend. I decided to go back into robotics while earning out the acquisition and worked on scenarios and simulation at Lyft Level 5, their autonomous driving division.

In late 2019, I started angel investing full time. Those early investments are rolled into Tango.vc Fund I, started in the middle of 2020.

Track Record: Cruise Automation & Powerlaws

When talking about track record, it’s hard to overstate the degree to which returns look like a powerlaw. This means a few top investments determine the whole fund outcome.

My very first angel investment was in Cruise Automation. It returned 62X, turning $25K into $1.5M. This was in just a few years, representing 294% IRR. If I had 19 other failed angel investments that literally netted $0, the overall return would still outperform most funds.1

The fund has made around 20 investments, which you can see below. At the seed stage, it is very hard to predict what might happen long term. That said, it’s worth exploring this first investment to understand how I think and work. 

At iRobot in 2007, I worked on the DARPA Urban Grand Challenge in 2007 and on adding better DSPs for video processing to bomb disposal robots. When Justin.tv was announced, I got in touch with the team. This was one of my first connections to Justin Kan, Kyle Vogt, and the larger YC community. It helped me push to leave iRobot, found my first startup, and go through YC in the next batch, W08.

When you’re an expert in a field but don’t work on it for a while, ideas accumulate. At Facebook and Dropbox, I learned what high scale web systems look like. This combined into a novel strategy of scaling training data for autonomous driving. I was telling my friend Aydin Senkut from Felicis Ventures my ideas in 2013, and he told me Kyle was working on a similar approach at Cruise. (Thanks Aydin!)

I met up with Kyle at The Creamery and asked to invest the next day. Keep in mind this is before YC, when he had a few interns and some 3D printers. This access matters: the last investor joining at Demo Day would have 80% lower returns, despite also investing in the seed stage. Around Demo Day W14, I wrote publicly about my ideas, keeping Cruise’s name out of it. Here is that contemporary blog post, published mainly to help with recruiting (which worked).

The rest is history, with a strong Series A and an early, enormous exit to GM.

What are the lessons here? 

• Understand the tech to see the connection between video streaming on the web and high scale ML training

• Use your network to get access to these great opportunities

• Pounce on great opportunities when they present themselves

This is how I’m running Tango.vc.

What Founders Say

"Beyond just being a very clear and independent thinker, Ivan was constantly sending me highly qualified candidates during the early days of Cruise - exactly when I needed them most." Kyle Vogt - Cruise cofounder

“Ivan’s experience shows when you sit down with him and talk about your business. He gets to an understanding of the business quickly and then provides multiple contextually relevant ways he can help you succeed. He is the type of investor you want on your side.” Hiten Shah - FYI cofounder

"Ivan has been a dream investor for Zitara.  He's generous and proactive with useful connections to clients and industry players, and has quickly earned our trust as a thoughtful and experienced business and product thinker with a uniquely clear understanding of ML technology development.  Ivan is one of our go-to investors as founders when we're looking for honest feedback on real problems." Shyam Srinivasan - Zitara cofounder

"Apart from the investment, Ivan takes time to deeply understand the product and brings pragmatic go-to-market ideas to the table. His network has helped us a lot in business intros, backchannels with investors and angel connects. If you're building developer tools, highly recommend getting the initial check from him." Jai Pradeesh - DeepSource cofounder

Fund structure

Our Fund I target is just under $10M with 20% carry, and 2% fee average, front-loaded. Currently, our check size to startups is around $100K. This smaller check size is very agile in co-investing with other leads, compared to writing $1M checks and competing with other leads. 

Because we’re targeting under $10M, there is a 250 investor limit. Previously, we were bumping the minimums as we got more commitments to manage this limit. Now the minimum will stay $25K for individuals and $250K for institutions and family offices. See more in this updated post about the change.

We have $3.8M committed with 62 LPs.

For funds & family offices: $250K minimum https://angel.co/i/aPEzB

For individuals: $25K minimum https://angel.co/i/xXmYi

We held an open LP Q&A, and here are the notes.

My anchor LPs helped me raise enough to start deploying quickly. I think it’s an incredible signal of support when my past investors in my startup back this new fund. This includes Roy Bahat from Bloomberg Beta, Trevor Blackwell from Y Combinator, and Lee Hower from NextView ventures.

Because we had LP commitments from my past investors right at the start, we’ve been raising and deploying in parallel. I set 3 capital calls to make it even easier for friends to access. We just had our 2nd, so when an LP commits, it means wiring ⅔. The last ⅓ will be called in H2 2021.

Companies & Performance 

• (Stealth) - Resilient hosting 

• Albedo - high res satellite imagery

• Almanac - Docs & workflows for async collaboration.

• Climate Robotics - Robots to fight climate change

• DeepSource - Automated code reviews

• Depict.ai - Automatic product recommendations

• Edia - AI for math education 

• Faction - driverless delivery

• Fathom Computing - 10,000X performance AI chips (exit)

• FYI - Search docs and manage access 

• Freshpaint - Heap autotrack with Segment syndication

• Hypotenuse.ai - Automatic product descriptions

• Locomation - Autonomous trucking

• Quartz - Computer Vision for construction sites

• Red Leader Tech - HD Lidar

• Rhetoric - Spellcheck for speaking (incubation)

• Stoke Space - Sprinter vans for Space

• Teleo - Heavy construction vehicle teleoperation

• Zitara - Enterprise battery analytics

Because this is a public post, I can’t release as many details about individual companies and their recent performance. I invested at the seed stage for each of them within the last 18 months. 3 have raised on SAFEs at higher caps representing 42%, 52%, 205% increase in enterprise value. Fathom was acquired. The companies want the details here outside a public blog post, but I can share more in a conference call. 👉 https://airtable.com/shrQnGGXbSbJ0zDOH

Tango.vc incubated Rhetoric and has common stock, making it special. The enterprise value created is already more than 3X the $1.1M we’ve deployed, which highlights how exceptional incubation economics can be. A future blog post will go into more detail about this math.

Coinvestors

Across all our seed investments, here are some of our coinvestor angels, funds, and individuals.

Thesis

Automation accelerates progress, which helps build enormous defensible businesses. We believe investors need to deeply understand technology in order to predict which will power those companies. This technical credibility helps win deals, with founders seeking investors who understand their products to help inform strategy.

Here are some of the patterns that get us excited.

Computers Can See

I first studied Computer Vision as a specialization in Robotics in 2003. Notable about that time: nothing worked! Since then, there has been a revolution with Deep Learning where tracking, classification, and segmentation are all very effective. There are very many verticals where using computer vision will enable new capabilities. Consider robots that don’t interact with an environment, but need to localize and move from A to B without hitting obstacles. This includes all AVs, delivery robots, and most drones. Solving computer vision means all these applications are now possible.

Computers Can’t Read… Yet

Can a computer analyze this paragraph and answer this question? Probably. Can it understand the context and infer related meaning? Probably not. There is currently a renaissance in natural language processing which enables whole new applications, stemming from the same progress in deep learning that helped computer vision. But text and conversation still require a great deal of common sense reasoning and knowledge. This means some new applications will be possible, while others will only become available years from now. A good example is marketing, where so much of the work is writing: ad copy, blogging, landing page copy, support email & chat, a knowledge base, and more. I soon expect to see teams of humans and machines, also known as “centaurs”. 

Computers Can’t Touch… Yet

Manipulation is the part of robotics where the robotic systems interact with the world. Think about your vacuum cleaner bristles that get stuck with hair, the clogged drain in your dishwasher, or the paperjam in your printer. These systems are expensive to develop and prone to errors. It’s no coincidence that the robots we most commonly see don’t actually touch their environment. An autonomous vehicle’s interaction with the world is just the tires. A drone has its rotors. In both of these cases, the robot needs to move from A to B safely and efficiently (a perception problem), but can’t do much when it gets there. This will change, and when it does, the application space for robotics will explode.

Machine Learning Dev is Still Hard

There is an enormous difference between research risk and execution risk. It used to be the question of machine learning systems was whether something might work or not at all. Now enough research and system development is done so that execution risk dominates. This means that the tools we use to build machine learning systems matter more than ever. But it’s still too hard. Analogously, databases went from rolling your own with specialized staff to standard open source solutions with a cornucopia of tools to help. We expect ML tools to follow this pattern: prototyping, collaboration, deployment, and monitoring. Some tools will be new, like managing training data or finding subtle errors in live models.

Vertical vs Horizontal

New technology can unlock multiple markets. Startups could take on a broad area directly, or sell to developers at other companies. It could also be that a new technology needs to be paired with customer and market insight, where a vertical deployment will win the space. There isn’t a general answer but it puts the other technologies in context. Maybe it isn’t a new robotic hand, but a custom robot to process recycling. Maybe it isn’t a developer-facing API, but an end application service enabled by the same technology. 

The Future

There is so much more to say. Here are working titles for future posts, and I’ll link from here when ready. 

• Fundamental math behind VC funds

• Why funds never hire, and how that affects diversity in VC

• How Tango.vc Fund II will be different

• Why management fees should be spent on software and services.

• How Tango.vc will fill the gap in seed stage recruiting

• The math behind the incubation model for VCs

Back the fund: https://angel.co/i/aPEzB

Want to learn more? https://airtable.com/shrQnGGXbSbJ0zDOH

Subscribe to get future updates.

Subscribe now

Share

Varda builds factories in space. Dalian is also a principal at Founders Fund who invested in SpaceX. Delian is now moonlighting as both a founder and a VC, which sounds really hard, but it turns out that folks at Founders Fund have done it quite a few times.

We talk about what that's like and:

• How you can iterate quickly on hardware

• Why we should building factories in space

• How you can build an early stellar team 

• How we should get more people working on ambitious startups. 

Here is the video, and there is a full transcript below.

Subscribe now

You have a new company called Varda and it's really ambitious. What are you trying to build?

I've been thinking about orbital manufacturing for quite some time. It’s related to how California became California. It wasn’t paying Lewis and Clark to explore. We became California via the gold rush.

Low earth orbit manufacturing is the equivalent for space where you can actually significantly accelerate, humanity’s path into the stars by commercializing lower earth orbit. This is as opposed to taking a big spaceship to Mars. That was the SciFi geek motivation behind starting the company.

The commercial reason to start the company is that there's actually a whole wide array of products that could significantly improve the quality of life for people here on earth that can only be manufactured in microgravity. People think making factories in space sounds crazy. But life here on earth would get better, like cheaper data transmission, a greater supply of human organs, much cheaper, cancer drugs, etc

There's a wide array of products that can only be manufactured up in space that are useful for people down here on earth. Nobody was solving that problem. I was trying to find a group to invest in and couldn't really find anybody. And so got frustrated and thought “Crap. I just got to do this myself or else it's never going to happen.” 

You are an investor, and now you're starting a new company. This keeps you very busy. How did you decide to do both vs one full time?

I've mentally tortured myself for years over, even before starting this company: what do I want to do? Do I want to go back and be a founder? if you asked 19-year-old Delian, I was consistent: I wanted to be a founder. I want to be a founding CEO. 

That's what I did. I dropped out of school at 19 and started my first company. Then I took a year-long break working at Teespring. The plan was to rest up and start another company.

But I couldn't find the right founding team for the company that I wanted to start. Keith Rabois offered for me to join Khosla Ventures. The role was somewhere between an EIR and Chief of Staff. I could learn from him and spend some time finding cofounders for my next startup.

Then I had an aha moment about eight months into Khosla Ventures. I realized first Venture Capital is the way to do space immediately rather than doing what Bezos and Musk did: making it big in normal tech and funding space companies. I can start funding these space companies today and work on space today. But only as a VC because I thought I wasn’t ready to found a space company myself at the time.

The second aha moment was that I kept meeting all these founders that are just like much better founders and CEOs than I ever was. I don't think I'm ever going to be a top 1% CEO. 

It comes down to personality traits and working style. I am extremely intellectually curious, ADD, bouncing off the walls, don’t care much for people’s feelings. Those are all huge pros when you're an investor. And those are all big cons if you want to be a really great CEO. I can work the 15 hour days when needed, but I would burn out a lot faster than great CEOs if I needed to do that for a long time. 

So at 8 months in, I thought I could credibly be in the top 1% of VCs. It might take some time, but I have a better shot than being in the top 1% of CEOs. So maybe I should stick to venture longer term. 

For Varda’s incubation, Will Bruey is the right CEO for Varda. I am not. When you look at him operate and the people he’s recruiting - he’s like a recruiting machine. We’re up to 12 people full time, and we got the cash in the bank literally a week ago. These are extremely experienced engineers: spacecraft, thermal, mechanical. All the types of engineers we need to build a company like this are in his Rolodex.

I know a handful of SpaceX engineers but it would be slower. The CEO’s primary job is recruiting.

Sure, there is a bit of a hedge staying a VC. But if Varda fails, who is going to want to take money from a guy who started a company and failed. My neck is on the line either way. 

How do you split your time? Do you have to be in different places?

Thankfully I don't have to reinvent a lot of this stuff from scratch.

This is a lot easier doing this three years after Founders Fund incubated Anduril. I can go to our partner Trae Stephens and just copy everything:

• The company equity structure to 

• The seed round term sheet with Founders Fund getting common and preferred

• The same employee offer letters

Anduril’s COO Matt Grimm has been incredibly helpful getting everything set up.

I had a long conversation that I've had with Trae Stephens around what he's learned over the past three There are a few key lessons that he taught me. 

First, you have to be regularly at the incubation. You can’t do it totally remotely. We’re both based in SF with the company in LA. You can try to do it ad hoc, but he quickly realized it doesn’t work. The team went through mental gymnastics on when to schedule in-person time. So just be there every Friday. Some Friday’s might not have a lot of Anduril business, but at least he’s there. People know when they can schedule in-person time.

Second, don't try to segment the days. Both jobs require being very responsive. There's a time where a Founders Fund portfolio CEO just needs to hop on the phone that day and discuss a new term sheet. Same for Varda. So you can’t do a magical segmentation. 

I literally just have two sets of TODO lists, inboxes, and priorities. At Founders Fund I could get to inbox zero. Now that’s impossible. Now it’s: where can I allocate resources to prevent something from exploding? For the past few months, it's been things like making sure we have workman’s comp, an office, and desks, that we’re hiring the right people. Varda has taken up a lot of mental space. 

But now once we get those hires onboarded, they're taking a lot off my plate. We have an IT guy, a head of business development, and a de facto COO. I think at that point, it'll start to get a little bit more evenly balanced. I'll do what a VC does best. I’m a really high-level head of people. I'm great at strategy,  fundraising, recruiting. That will be most of my contribution to Varda. I'll jump into the various high-level customer conversations, but we have a guy on the ground that's a lot better at that than I am. He'll rope me in when needed.

What is the recruiting list coming up and where can I help there? Will Bruey, the CEO, is gonna be a lot better at recruiting our VP of engineering. When it comes time for us to have a CFO or a CRO, I can significantly help accelerate those types of recruiting campaigns.

They intermingle. I'll be on an FF call one time. I'll be out Varda call later. There are also the fun ones that are actually both, where I'm talking to a space company that Varda can use and FF might want to fund. 

Even this interview! People are learning about investing and also about Varda. 

For Varda, you haven’t shared the details and timeline for what you’re building. What relationships do you expect? Maybe you have a pharma company that wants to build something and contract with you to build it. Maybe you build the robot that can manufacture that material. Or maybe the most profitable path is to be the only supplier of long carbon nanotubes you sell to others. 

What do you expect to happen from Seed to A over the next year or two?

We'd like to be more abstracted and just be the “Foxconn of space”, not developing or even designing the product. We're just a step in the supply chain with the most expertise in how to develop it. 

Early on, we have to develop that market and use cases ourselves. We're much more involved in the actual, basic design of the first products. And then we’re also leading those customer conversations.

Ideally in the long term, when we're starting to work on cancer drugs, Pfizer comes to us and says, this particular drug would benefit from microgravity manufacturing. Here’s the design, manufacture this for us, and handle the supply chain. We deal with microgravity manufacturing, bringing things back down, and delivering it to the customer.

With Foxconn, their “supplier” is Apple, where they buy designs. But then they are selling the iPhones right back to Apple. That’s the ideal outcome. 

For the first couple of products, it's not going to look like that. We’ll need to make the raw materials ourselves and also figure out the customer side. Josh Wolfe who joined the board as MD from Lux Capital said this first product is kind of like training wheels. The point is not to develop that product. The point is to develop that whole supply chain and then be able to apply that supply chain to a lot of different products.

It doesn’t matter if that first product is profitable or perfect. Generate some revenue so you can show it’s commercially viable. 

When it comes to venture for super deep tech, often by Series A or even B, companies often don’t have any revenue. But for a B2B SaaS product, you're definitely judged by retention, revenue, and other metrics. Where do you think it will be on that spectrum? Those training wheels can take you down the driveway or down the block. 

My presumption is that you're gonna have to raise a lot of money. The market is completely new, so maybe you need to prove the market exists. Or maybe you just need to prove the technology works. What do you expect? 

Thankfully space factories, as crazy as they sound, are actually a lot less capital intensive than rockets. Depending on the quality of that revenue from that initial product, it’s possible we break even on the R&D with $100M-$150M raised. 

We won’t need to raise $500M+ in order to get a significant amount of revenue. Broadly, we’re a lot closer to Relativity Space, raising lots of money before you have revenue. The promise is if this deep tech works, you get a binary outcome. If it works, it’s extremely valuable because nobody else can do it.

This depends on the investor community appetite for believing something like this. The reason that Relativity Spaces of the world are able to raise so much is that SpaceX is showing that the launch market is quite valuable. 

We don't have that same analogy in microgravity manufacturing. We need to be able to prove that story. Three or four months ago, I thought it would be pretty tough. I’m kind of crazy. Maybe I was a little bit too pessimistic. It turns out a lot of people have been thinking about this.

Maybe it’s easier to understand once we actually give a full pitch, and walk people through explaining the plans. There was a lot more appetite for people investing in Varda than I expected. We also had a lot of interest to back with very large checks from people interested in seeing this come to fruition.

We've put together the world-class team across the various technical and non-technical aspects of the company. It's our ball to fumble. I think investors will continue to be interested irrespective of the exact revenue metrics of the company through  D or E.

I think this company gets to $500M to $750M total raised until some real scrutiny starts to get put on the margins for these products that you're developing and selling back down on earth.

When it comes to manufacturing, there are really different kinds of robotics. Self-driving cars and drones are field robotics. What you see in manufacturing robotics is different. Even terms like computer vision vs machine vision - only in manufacturing do you talk machine vision. You have a controlled environment, so my presumption is it's more of the latter. But maybe it’s teleoperation. 

What kind of automation do you expect?

It’ll be largely automated. 

It’s funny, the first question I get is if its humans out there. Oh my god, no! That requires so much engineering. The ISS is not a simple thing to build where you have continuous oxygen and dealing with all the needs of humans.

And then people ask, will be like Boston Dynamics robots walking around a floor and picking things up? Oh my god, no! 

When we talk about factories, there's four motors. That's about it. Those motors spin at various rates and various things that get moved very simply. You do have to be very precise. These are high-value materials that are somewhat sensitive. But this is not robotic arms moving around. You move a pulley, spray a coating, or heat things up. These are extremely simple electrical and mechanical processes that were automating.

This is much simpler than a Tesla manufacturing line. It's not like the entire product is manufactured in space. It's the very particular step that is sensitive to being in space. Most of the time we have an almost done product and we’re finishing in space. Or you're taking some pretty raw materials, doing one step in the process, and bringing it down to earth. The rest gets done on earth. 

It'd be fun to have an orbital dropship that lands on my front yard with the iPhone inside. But probably not the way it's gonna work. 

For what it's worth, not that far away though! 

The thing about these markets is they're just so, so large. If we make it work, with resupplying our factory continuously just with materials, even automated, we'll consume more launches than all the satellite companies combined. We’ll double the entire launch market within a couple of years. 

I don't think it's impossible that within a decade, we will be producing more semiconductors in space than we are on earth. And Varda will be the primary producers of those. An iPhone is a set of semiconductors welded together. It wouldn't be impossible to see an entire Apple manufacturing line in orbit within a decade. 

In planning this company, did you take the launch costs curve and just project it down? What do you expect and what is required? And will this be more SpaceX launches where the costs come down, or is there some other dimension?

On SpaceX’s site, you can book a rideshare for a 200 kg satellite for a million dollars. That’s $5K/kg to launch. In the projections we pitched to investors, we assume by 2025 that goes down to like $3K/kg.

I talk to friends working on Starship, like my fraternity brother from MIT is head of economic performance and responsible for figuring out how much it’s going to cost to launch. He's always delivered on all his various promises. He's been in SpaceX for about nine years. Sometimes they're a year or two late, but they always deliver on their promises. He's promised me $50-$100/kg by 2025. That makes microgravity manufacturing move from difficult to make profitable to extremely easy. 

Everybody's mind needs to shift. It costs you a million dollars to send your satellite to space today, and you don’t spend less than $1M designing it. If you spent $50K making it and it fails, you lost all this money in the launch. 

But if it costs $50K to launch, then all of a sudden you're talking about satellites at a 50K. Our ability to send an ISS size module up to space as a non-government non-state entity in that world. I'm not banking entirely on SpaceX. Look at the number of  launch companies that are both getting into orbit and are getting significant fundraising rounds. They are also thinking about the economic cost of this and our margin down towards more and more reusability. We now have three commercial companies that have made it orbit: SpaceX, Rocketlab, and now Astra as of like three weeks ago. It’ll be 6-7 in the next 18 months. All of those companies are increasing production and slashing costs and improving the economics.

Even if Elon gets hit by a bus and Starship doesn't work out, we're already on this decreasing cost curve. It is moving far faster than Moore's law. There's no fundamental physics reason that it should stop. 

We project $3,000 a kilogram by 2025, but I think it will be much lower than that. I think a lot more people will be thinking, damn, why didn’t I start a microgravity manufacturing company in 2020? 

One of my favorite topics is moving fast and hardware, because it's just so hard. The way you make something cheap is that you lock in a design for manufacturing. The way you make something good is iterate. I know Founders Fund has strong opinions about having a clear vision for the future. Go and build that vision, versus going “lean startup”. If you have a $50K versus $5M satellite, how much work do you put into that? Look at the way NASA does a Mars mission. You have to be very conservative. So how do you build a company that can build hardware fast? 

The average tenure at SpaceX of an employee at Varda is about 6.5 years.

Thankfully SpaceX knows how to make hardware really, really fast. We already have that culture and are closer to the “lean startup” model. We create hardware iterations rather than try to design the end product.

Starship is now the perfect example of this: nonstop creating iterations. Don't increase the technical scope too much because if you fail you don’t know the root cause. We're starting off with extreme MVPs of our basically space factories. We make sure they work and then add on scope to eventually get ready for launch into orbit. 

I definitely don't believe in the lean startup model when for, “What you are building in the long term? What is your business?” and AB testing towards that.

When it comes to hardware development and space companies in particular, they get screwed when they try to come up with a design on a whiteboard and then perfectly designed towards that. That’s what prevented blue origin from getting to orbit. Plenty of other space companies suffered the same downfall. 

You have to focus on making a piece of hardware every quarter, testing the full system, and integrating together. Then every team does back: the re-entry team makes the next version, the microgravity manufacturing team makes their next version, the solar panels, and propulsion. They reconvene every 3 months and integrate everything together and see how it works. This or that failed, now we know what we need to fix for the next one. 

If anything, I'm concerned about this team moving too fast given that speed is in the DNA. 

We went from founding team, to capital in the bank, to the first 12 hires, faster than any space company ever. So let's make sure not to go to orbit too soon. Our timelines are only getting accelerated. We thought our electric engineer wasn't going to start for another 3 months. They're starting on day one in the office of everybody else, so we can start electrical design a little bit faster than we expected. 

It's incredible to see the recent Starship test that exploded. But everyone cheered it on because they learned what they needed to learn. Only a company that's focusing on the iteration can take the value of that.

I’ve meaning to tweet about that particular Starship analogy how things have shifted. My buddy high up at Starship told me how it’s bending their brains at SpaceX because they've had to all unlearn “rockets exploding is bad”. For a long time at SpaceX, a rocket exploding almost killed the company many times. Elon had forced to team to launch, regardless of how ready you are. You’re launching today because we’re in a post-capital world at SpaceX. Rockets blowing up doesn't matter anymore. I can afford to build a hundred of these before we run out of capital. 

We're not gonna run out of capital. It's really amazing to see that ethos come to fruition. You also see the public start to appreciate this as well. Read books like Loonshots or other books about mega projects and how they work. It's constant hardware iteration and willingness to have that hardware fail. Designing in a vacuum, on a spreadsheet, and on a whiteboard never works.

Just send things out in the environment on a regular basis. As a space enthusiast and entrepreneur, it is thrilling to see that the public markets and private markets are willing to make space companies live in a “post capital world”. Elon can raise infinite amounts of money for SpaceX. 

All that SpaceX has done over the past decade has been in a pre-capital world. They had relatively limited fundraisings, and some were really tough. They have raised more in the past 12 months than they did in the entire company's history before that. If you think SpaceX is cool, imagine what they’ll do in a post capital world. 

People joke about late-stage capitalism, but actually, it means getting to space, becoming a multi-planetary species, and watching cool rockets blast off. 

How do you get founders to be more ambitious? Maybe it’s just by demonstration. Let's look long-term. What does the future you want to live in? But people tend to hedge. To caricature, every employee at FAANG believes they should be focusing on short-term income over building the future. How do you get all of these would-be founders and people starting today to dial it up, to build the world we want? 

The way that you encourage us is actually just by practice. The reason that I was more willing to start Varda was that I was seeing other financing rounds happen. I thought, Woah, you can raise a lot more with the level of team that I could recruit than what you could 4 or 5 years ago. 

Varda couldn't have raised a $9 million seed round out of the gate five years ago. I was in the fundraising ecosystem as an entrepreneur in 2015. Zero chance. The only people that got $9 million rounds out of the gate were the Jack Dorsey's of the world. I'm no Jack Dorsey, but it's pretty awesome that I can get $9 million for a pretty damn crazy idea. I do think that type of stuff really does encourage the next generation of entrepreneurs. 

Second, I do actually think that this is starting to shift. I'm more optimistic about company formation than I have been in a while. If you just look at macro US stats, more new businesses were formed in 2020 in the United States than any year in the past decade.

The oversupply of capital is now starting to shift people's mindsets. If you were an early employee at Stripe or the equivalent, you maybe got financed, but it wasn't like a guarantee. Today, if you're an early employee at Stripe, Anduril, Snowflake etc, it almost doesn't matter what your idea is. You get to raise three or $4 million out of the gate. The bar has been lowered for both quality of team and quality of idea. Back in 2009, $3M or $4M would have been called the Series A.

It's increasing the success rate of these companies too. Way more capital's afforded to them that allows them to continue to progress. The quality of startup advice, mentors, and number of VCs continue to increase since I've been in Silicon Valley.

You keep inspiring people by doing and being the example. I am sure that a lot of my friends are thinking, Delian wasn't 10 years at SpaceX, and he’s starting a space company. Maybe I can also go start a space company! 

It used to be space and Silicon Valley were two separate areas. The overlap between the two worlds is getting a lot tighter. Julia DeWahl used to be the chief of staff at Opendoor and now is one of the early business hires at Starlink. I categorized her into a Silicon Valley world. I never thought she would overlap in a space world. Worlds are starting to bleed together. I think that gets people a lot more comfortable working on space. They’re thinking about the kinds of companies they can now start. 

Subscribe now

See our interview here, with a full transcript below.

Tell me about DeepSource. What are you building?‍

At DeepSource, we are building an automated code review tool that helps developers find and fix issues in their code.

The grand mission is helping people write better code using automation. We automate things in the software development life cycle.

Look back to how we write English, for example. If you have access to the internet, spell checkers, and grammar checkers, it’s much easier to write English. 

But the same thing has not really happened for code. Developers have IDEs, but IDEs stop at style detection and rudimentary factoring. Computers should be able to help you write better code. 

‍

This is very high leverage activity because software engineering is like an engine of productivity. If you make the engine faster and more efficient, then all the cars drive faster. I get really excited for tools that help people make tools faster and help develop our technology. 

Can you tell me a little bit about the culture within deep source? What are you trying to build internally?‍

We have a culture that stems from what we are building.

We have been fortunate enough to have people in the team who really care about building these tools that other developers use. Actually very few companies in the world work in this space. Seeking out people who care about this is also a big part of what we do.

Internally, we see how things have been done, how other tools or other companies have tried to solve the same problem. But also we look at it with fresh eyes: how can developers take action from our product?

We’re focused on the user experience using a lot of data. How many issues are people fixing using deep source? Which issues raised are people fixing or not fixing? What explains the user behavior? We have a very active Slack group with a lot of our early power users. We talk to them to figure out what to build so developers use the product every day. ‍

‍

There is this authenticity that comes from people that care about their craft and what they're trying to build. It also comes in the marketing side, where developers hate being sold. They just love the people that are actually in the trenches helping build with them.

You’re looking to hire a head of marketing. Describe what you’re looking for.‍

It’s absolutely true: developers hate being sold to. You can’t shove a product onto users using traditional tactics. So we’re looking to get DeepSource in front of more people to show them how they can save time. At least 40 to 50% of their work doing code reviews is now automated using DeepSource. We do this without obstructing your existing workflow. We integrate seamlessly with your existing workflow.

Developers love automating stuff. They’re passionate about automating things that can be automated.

That's what we're looking for in our marketing. We are hiring a Director of Marketing who can understand and empathize with the problem that we are solving. We want someone who can help us position the product in front of millions of developers. ‍

‍

There an interesting combination of working with data and sincerity that comes with your audience. I see this as a great opportunity because there's definitely a lot of quantitative marketing even for building community, even when you are using different tools than just, you know, some ad campaign or building out leads for our sales team.

Do you have a sales team or is this all self-serve? How would marketing work with different sales roles? ‍

We don't have a sales team at the moment. All our customers have been inbound. 

The product is free for open source and that is the primary driver of our growth. Developers try us on open source projects they are working on. Most of them work at a private company and start using DeepSource on their private code repos. 

We actually have not spent anything on marketing as well. We still want to figure out what else to try in marketing? This role is going to be exciting because they get a chance to both drive what is working and explore new areas. We have a blog that is working: our content on static analysis, code reviews, writing good code, etc bring us a lot of traffic. 

The primary challenge is how to position ourselves? 

To learn more about the role, go here. 
Or email Sanket 

I think there are thousands of engineers in his position, without a strong background in sales and considering how to step on the gas. So I asked Adam Gross to join us. I met Adam early at Dropbox where he lead Sales. He has also invested in multiple developer-facing companies and served as CEO of Heroku. 

This is a masterclass crash course on how to think about getting started in sales.

Cronitor started out of a need to monitor the status of dozens of background jobs. It became a side project and grew slowly. Most of the customers come from SEO through cron based content marketing.

Now they have $28K MRR with zero outbound and no sales focus. What should be the very next step? The goals this year are to figure out how to grow faster. This is obviously a common question for many developers. 

Adam’s first question was about scale, with that MRR and 500 customers, with a few thousand users, growing revenue around 5% MoM. On the technical side, Cronitor ingests 30M pings days of telemetry. 

The 2nd question is the business goal. August wants to understand what to do beyond self serve. 

There are three categories:

• How to acquire users?

• How to move people through the funnel?

• How do you increase average contract value? 

When you have a lot of VC money, you tend to build out a marketing function just because you can. Cronitor is bootstrapped to date, so is investing in marketing with more care, but many more companies should carefully consider what to do.

Doing more outbound means increasing the deal size because of the cost of a sales function. But Cronitor at $2500/year can’t support that kind of sales infrastructure. 

What is the rule of thumb? A sales rep should bring in $1M/year. For $2500 per deal, that would be a very high pace for a sales rep. Using the LTV is better than the first year revenue, but this is way off. 

So what should Cronitor do with it’s prices? Cronitor is a bottoms-up developer-first GTM. There are 3 modes that work:

• Free

• Self serve paid

• Enterprise

Typically, you build the free channel, then the self-serve, and then enterprise. Adam’s rule of thumb is don’t build enterprise until you’re $10M ARR. That might seem very high! But the cost of building out enterprise sales and support is very high, and Cronitor has so much more to grow with self serve.

One reason is that you need a large funnel of free and self serve customers to support the enterprise pipeline. You’ll work to expand those free and self serve channels anyway to try to make enterprise work. But it’s also the organizational cost. 

It’s like a meal kit vs a restaurant. There is a whole organization required in a restaurant to serve a meal. Meal kit is self serve and really different. 

The Crontitor pricing page highlights the trial when there is actually a free tier to get started. 

You could also consider selling the sizzle. My favorite analogy is from User Onboard  where you’re selling Mario with fireballs, not the power flower.

So what do you invest in if not in sales? Support! Hiring an excellent support rep will help those who are already coming to your service and consider buying. Support is marketing and docs are marketing. 

Beyond that, examine your users to aggregate different users across a large organization and then sell to that org. 

There is also refinement on pricing for Cronitor. A very small percentage of people are on the “Startup” plan at $49/mo, which means ops isn’t nearly as price sensitive. Experimenting here is hard, but it would make more sense to let people start free and then charge more for the teams that do integrate. 

But how to change prices? Adam has an incredibly valuable idea: grandfather in users for a time, but couple the change with more features. Only allow people to get the new features if they step out of the grandfathered pricing. This is a way of incentivizing upgrading pricing, and users will think it’s a fair trade. 

What about fundraising? There are rules of thumb like hitting an ARR number to raise a seed, or what might get you into Y Combinator and hitting demo day hard. But first consider whether you want to fundraise, which would mean you’re aligned with the venture model.

Assuming you want to raise, what counts for first-money-in fundraising is adoption and market, not revenue. An investor is asking “how big can this be?” and not “how do I get people to spend $49/mo”. Then step 1 is how you acquire people, which is SEO today. 

As an investor, Adam loves to look at community affinity. Heroku was rails. Github was git. What might that community be for Cronitor? Maybe something around databases and backup?

Subscribe now

The result is this video which goes into lots of details. Please subscribe on YouTube. I put a timed transcript below, along with the notes I used to prepare for the session.

Subscribe now

0:00 Introduction

‍0:26 What is "Not Boring"?

‍2:21 Who is your audience?

‍3:11 The goal is growth, while retaining quality.

‍4:35 The economics of syndicates vs fund

‍7:08 How to configure content for premium vs free subscribers

‍9:45 Exploring the new DIY media trends

‍11:30 Writing about small vs large companies

‍12:22 What tools do you use for your Substack?

‍14:25 Webflow vs other landing page tools like Landen.co

‍15:50 Ivan's stack at Tango.vc

‍16:55 How does Growsurf help?

‍19:20 Start diving into the user flow

‍19:55 Guessing Hunter Walk's name on the leaderboard

‍20:30 Powerlaw distribution of the referrer leaderboard.

‍21:15 Launching your Substack on Product Hunt

‍22:55 Ivan Kirigin's background helping run growth at Dropbox and founding YesGraph to help startups grow

‍25:45 Getting subscribers vs getting referrers, and optimizing the related call to action

‍27:15 What to send in a welcome email? Too many links is confusing.

‍28:40 Feature request: AB testing calls to action in Substack welcome emails. How Dropbox ran AB tests for referrals

‍31:00 How stickers and t-shirts are surprisingly motivating.

‍31:25 Don't make the user do more work. For example, if you already know who they are, don't ask for their email again

‍32:35 Why the referral sharing channel matters. You could break down successful shares across channels like Facebook, Dropbox, and Twitter

‍34:45 The challenge of writing copy that will put words in a user's mouth

‍36:15 Why importing contacts and send-all is bad. Ivan shares a story from Dropbox about removing send-all.

‍38:00 Personalizing landing pages helps improve conversion.

‍39:25 How social proof improves landing page conversion.

‍40:15 How to break down startup metrics.

‍42:45 Why to measure viral loops continuously not in average aggregate.

‍44:45 How Substack building their own referral features would help the ease of the flow.

‍46:45 How tracking invites helps optimize the flow.

‍47:15 Why the biggest mistake is not promoting the referral program enough.

‍48:45 What should Substack put in a referral program? After including a referral link, earning a paid subscription would be incredible.

‍50:40 Allowing you to test which posts bring in the the most new users.

‍51:50 How good content matters for the amount of growth, with Y Combinator's YouTube channel used as an example. 1/3 of all their traffic is from 3 videos about Elon Musk and Mark Zuckerberg.

Subscribe now

Profile Link (or tweet, or elsewhere)

Two links: domain and substack.

Substack landing page

Obvious goal: subscribe

Individual Post Landing Page 

no links to referrals

Subscribe Email

link to referrals, with 6 other links

Referral sending page

• Slow to load

• No way to participate above the fold, but obvious to scroll

• Not obvious those are stickers to win

Enter email form to participate

This is an extra step. Many already have given you the email, you’re just not saving it.

Referral message

• Nice subject

• Clear call to action

• Editable by user, without the UI

Referral landing page == subscribe page

• Who is the inviter?

• Subject works no matter who invited

• Product hunt general social proof, along with “6000+”

Metrics:

These are for each time period, meaning they are a loop

• On referral program / all subscribers # put links in newsletter - or their referral link

• Invites / inviter # is this transparent in GrowFlow? Your analytics? Eg URL grsf=*

• New / invites #

Specific Metrics

Track total referrals right before each newsletter is sent.

Put each in a spreadsheet, with (date, total, new).

Include total subscribers to get 

• aggregate growth rate: new subscribers / total subscribers

• total referral growth rate: new subscribers from referral / total subscribers 

• percent of growth from referrals: new subscribers from referral / new subscribers

Interpolate if you need to get standard dates, eg weekly referral growth

Run longitudinal AB tests on (positioning, copytext, color) of referral prompts

• If total referral growth rate increases, good

• Even if percent of growth from referrals increases: good

• Caveat: longitudinal tests will have conflating variables. Examples:

• Marginal subscribers might default to lower referrer rates. 

• Marginal potential audience might be less eager to subscribe, lowering all channels

• Your content varies!

Subscribe now

• Apple has no product managers! The "Directly Responsible Individual" is on point to deliver, and it doesn't predict your position in the org chart.

• What happens when your project gets defunded? People get worried, like they're getting kicked out of the tribe. Actually they should have been worried about working on something that didn't matter.

• Key to an organization is sustainable coupling points. "How could I set up my team so that they could ship in a year all without VPN access?"

• In the Harvard Business Review in 200 years, what are the things that they'll say? It's illegal to have an organization larger than a certain size like Dunbar's number?"

• "Hey Siri, terraform mars" - what would it take to make that work?

• We made a team like the SEC who collected metrics and called bullshit about any team at Apple.

• I see growth teams looking to take their product roadmap, break out factors like impact, cost, risk, and want to add it up to a single score and just do that. But there is more judgment required.

• How important is a leader required to explain the rationalization behind a decision? It's costly and sometimes impossible.

• How does a leader justify a no-hire decision based on a gut read that the person just doesn't fit the org?

  Subscribe now