Why We Stopped Working on Athens, and What I'm Doing Next

2 years, 9 months, and $850k later

Why We Stopped Working on Athens

In November, I announced that we were winding down the open-source project I founded, Athens Research.

Jeff Tang 🏛 (LA) @tangjeff0

Athens the OSS project is winding down. The company is still operating, but taking time to reset and explore new ideas. Open to chats and convos. Thanks all ❤️

Athens 🏛 @AthensResearch

Update from our Discord: https://t.co/EdDvO20Dr2

These are the reasons I made this decision.

• I no longer believed there was a billion dollar opportunity in note-taking, knowledge management, or knowledge graphs based on the customer conversations we had.

• I didn’t feel like I had any interesting ideas to contribute to the space anymore — Tana implemented many of the ideas I wanted to.

• The explosion of progress in AI made me wonder if AI or AI-native tools might leapfrog the current paradigm of note-taking and knowledge management.

One day, I might elaborate on these reasons and share more about what I learned building Athens. But we also have to keep moving. We still have money in the bank. People still believe in us. And there are still many great opportunities out there.

---

What I’m Doing Next

The purpose of this newsletter is to document the pivot. The rest of this post will be about what I’ve been up to since.

I’ve been exploring four broad themes:

1. Vertical SaaS

2. AI

3. Space Tech / Robotics / Manufacturing / Defense

4. Health / Biotech / Longevity / Neurotech

I’ve also been talking to a lot of founders, engineers, and writers to get inspired, to learn about other industries, and find a place to work if this doesn’t work out!

Here are the conversations I had in December and what I learned:

• YC Construction Tech Founder

  • Believes upskilling and training is an opportunity in labor-heavy industries.

  • Labor marketplaces are hard.

    • Supply has adverse selection. The best workers always have work, the bad workers use marketplaces.

    • Hard to make money. Low margins, make money based on bids.

  • Challenges with Vertical SaaS

    • Margins are bad in legacy industries.

    • The low-hanging fruit, SaaS 1.0 point solutions days are over.

    • It’s all execution risk. Just gotta grind out a ton of software and sales.

    • Hiring. Average Ivy League CS person doesn’t want to work in the space, but talent is one of the entire advantages of being a startup in a legacy industry.

    • This will never be cool. You should value building a company and solving a problem.

    • Have to be great at sales. Most tech people suck at sales.

  • Playbook for Vertical SaaS

    • Find a problem. Shouldn’t be hard, because there are many.

    • Figure out how bad it is. It should be a $50k-$100k problem. Essentially worth a person.

    • Commit to no feedback, no customers, and grind it out for a few months

      • You can’t launch a half solution for payroll. You can’t launch a half solution for trucking management. You need the whole thing.

      • If you have design partners, obviously that’s good.

    • Recruit young, entrepreneurial product people. Lots of full-stack eng.

• Wil Eyi, CEO of Toolbox (YC S20)

  • Wil basically applied YC’s and Airbyte’s advice on pivoting.

    • Pivoting doesn't need to feel like pivoting if you have a process. Also it shouldn't take too long to validate ideas: 2-4 weeks.

  • Grouping all Vertical SaaS's across different verticals or fintech applications is a gross over-simplification.

    • Credit is very different from banking or payroll.

    • Construction is very different from manufacturing.

  • If you have a big vision and a path to getting there, it doesn't matter how small your entrypoint is.

  • Space Tech and going multi-planetary is cool, but so is modernizing multi-trillion dollar industries.

  • Once you have nothing to lose, you are free to do what you want, free to discard VC aphorisms, and free to do what's right: solving customer problems.

  • You don’t need to have your whole life planned out or committed to one problem. Most founders are not committed to life to ONE area. Most founders have a lot of interests.

  • Also believes upskilling and training is an opportunity in labor-heavy industries.

• Wiley Jones, Co-founder of Odonata

  • Healthcare

    • All roads lead back to billing. You need actual bottom-line improvements for healthcare companies. Insert yourself in the billing loop. It is harder but much higher revenue wise.

    • If you do help with managing claims and medical coding, it is operationally difficult. It’s more of a tech-enabled company than a software company.

    • Selling to these companies is awful.

    • People will pay for financial tools, not productivity tools.

  • Vertical SaaS

    • There is a massive whitespace between email/spreadsheets and SAP/Netsuite/Oracle.

      • Email/spreadsheets — used by $5M companies

      • SAP/Netsuite/Oracle — used by $1B companies

      • Odonata vision: everything in between.

    • People hate SAP, Oracle, etc. because of how horrible and expensive they are. FTX, which was (fraudulently) handling billions of dollars, used QuickBooks instead of the enterprise tools.

• Kai Mckinney, Co-founder of Relay

  • Insights about the Ohio startup scene. Got me pretty excited about Columbus!

• Justin Olshavasky, Co-founder of Voyage Bio (YC S19)

  • Got me excited about epic startups in the biotech/medtech/health space!

    • Voyage makes a device that cools the brain down. This helps people who are having strokes. Normally, if you have a stroke, you have brain damage after 5 minutes due to lack of blood flow. Cooling slows down the blood flow.

    • They tested this on a pig, Wilbur. The blood flow to his brain was cut off for 30 minutes, which should’ve killed him in normal circumstances, but Wilbur had a total recovery after, eating, drinking and running around.

  • There is demand for good software in bio.

  • Perhaps using Rust for embedded systems could attract more developers into bio.

  • Software companies for health companies: Latch, Benchling

• Peeyush Shrivastava, CEO of Genetesis (Thiel Fellow ‘17)

  • Peeyush has been working on his MedTech startup for over 7 years now. And they are just now getting to their first 5 paid customers.

  • Cincinnati, Ohio is actually quite good for biotech, defense, and insurance companies.

    • Lots of hospitals and insurance companies here.

    • L3 Harris has an office in Cincinnati and Dayton has an Air Force base.

  • Believes the future of health is consumer-first.

• Dival Bannerjee, Co-founder of Vuecason

  • A lot of the complexity for manufacturing really comes from supply chains.

  • Manufacturing companies (SpaceX, Apple, Amazon, etc.) often do vertical integration out of necessity. You have to own your supply chain otherwise you expose yourself to a lot of risk.

  • Like Justin Olshavasky about medtech, thinks Rust for embedded languages would be cool.

  • There is demand for good software in hard tech and in LA.

  • Applying software frameworks/methods/practices to manufacturing could be powerful, e.g. CI/CD, linting, deploys, etc.

• Ry Walker, Co-founder of Astronomer (Raised $300M)

  • Sees every non-open-source company as being potentially disrupted by an open-source company. At the very least, the core should always be open.

• Jeff Burke, Biz Dev at Repl.it and writer

  • Consider working at a company first like Hadrian to get experience.

  • Software companies for hardware companies: Galvanick, Epsilon3.

  • Consider selling to space / defense / advancing manufacturing startups and sell to them, not the government itself, which takes forever.

  • Have to fundraise a lot for hard tech.

• Ahmed, Writer at Post Apathy

  • Not bullish on charter cities generally speaking, but if one succeeds it will be Praxis. Other charter cities are in South America or Africa, but you need a more stable government. Praxis will be in the Mediterranean, which is more stable, and arguably is the best land on the planet.

  • The single most important thing America can do to revamp itself is to improve its immigration policy.

  • All predictions made for 2022 were basically wrong. Predictions for 2023 will be tricky.

    • No one expected Russia to invade Ukraine.

    • No one expected Russia to do so poorly.

    • No one expected China to end up handling COVID so poorly, when they started off so strong.

---

This past week in LA, I spoke with:

• Chris Power, CEO of Hadrian (raised $100M)

  • All hard-tech companies are financial engineering and operational engineering companies.

  • No cofounders, because would’ve needed 7 technical cofounders for what Hadrian is doing.

  • Marketplaces for physical parts do not solve the underlying supply chain problems.

  • The factory is the product.

• Aaron Slodov, CEO of Atomic Industries

  • [Stealth]

  • Build the tool that builds the tools.

• Joshua Steinman, CEO of Galvanick

  • “You should build in public your pivot” → inspired this newsletter

  • “You should take 6-12 months to do proper customer discovery.”

  • Go to conferences, do the work, talk to customers.

  • Listen to Ari. If he says build a satellite imaging startup, do that.

• Edward Mehr, CEO of Machina Labs

  • The problem with traditional factories is that they are one-off. For instance, the factories for Falcon or Raptor are designed around the rocketship. It is easier to create a new factory than it is to repurpose an existing. So we have all these abandoned factories in the Rust Belt.

  • Machina Labs is inventing a new process for creating sheet metal — can think of it as 3D printing metal. Not only can they make some parts faster or cheaper, they can create entirely new metal sheets that were not possible before based on physics, heat, etc.

  • Factories today are like supercomputers built for one purpose. What if factories were more like data centers?

  • Hardtech / LA needs good software engineers.

• Ari Lewis, Co-founder of Payload Space

  • Don’t sell to startups that do space / defense / advancing manufacturing. It’s too early.

  • Create a satellite imagery / earth observation / change management startup.

    • Read Planet Labs case studies and Joe Morrison tweets and blogs.

    • Pick a vertical, go to conferences, and sell.

• Justin Lopas, Head of Manufacturing at Anduril

  • Building software for hardware is hard because in order to a great job of one, you have to understand both.

  • These manufacturing companies are often binary. You are either a mom and pop shop or a startup trying to be the next Raytheon, Boeing, Lockheed, etc.

• Brandon Kleinman, Co-founder of Birthday App, Showroom, and Laurel & Wolf

  • The defining characteristic of being a founder is agency. The tradeoff is loneliness.

  • It is ironically lonelier as the CEO of a 100-person company than a smaller startup. Furthermore, it is a battle to get people to tell you accurate information.

  • Listen to the Founders Podcast.

• Will Depue, Co-founder of MithrilHQ, Colossus

  • If you are a young ambitious builder, you need to be on Twitter. But no one tells you that.

  • We need to enable young people by getting them together.

• John Nabuurs, Engineer at Cover Home

  • The mission is a huge plus at Cover. Everyone gets it. It gets everyone to show up and push harder.

  • We need to enable affordable housing.

• Greg Castle, Investor at Anorak

  • Consider SBIR or AFWERX as programs to break into defense.

  • While defense/hard-tech companies might say that software is easy, a lot of them have weak points in the end-user product and UX.

  • Hard tech companies often led by PhDs and technology nerds. There is still a need for good products and experiences.

• Addam Jensen and Tristan Yang, Co-founders of Holocron

  • Having relationships in defense is super key.

  • Dual-use tech is good because when economy isn’t doing well, defense spending goes up, and when economy is doing well, commercial spending goes up.

---

Now, I’m back in San Francisco, where I’ll probably be until March. Later this month, I’ll be going to Scale’s AI hackathon and Orangewood Labs’s Hard Tech hackathon.

I will be posting updates here and on Twitter @tangjeff0.

Thanks to Dillon Zhang, David Phillips, June Li, and Erich Lehman for their feedback on this post. Thanks to my investors for betting on me, and their continued support. Thanks to all the people willing to have conversations with me. And thanks to the Athens Research community and anyone who ever supported or contributed to our project. Onwards.