The Lean Company Playbook

Jake Cooper left Uber in 2020 with a box of databases for personal projects, no job offers and a few angel checks.

Five years later, Cooper's startup Railway runs backends, frontends and AI apps for more than 2 million users and adds about 200,000 developers a month. Teams at 31% of the Fortune 500 use Railway, and it got there with no marketing for most of those years. Along the way it left Google Cloud, built its own data centers and now owns about $90 million of hardware, rare for a software company. In January 2026 it raised a $100 million Series B led by TQ Ventures.

The number that makes this a playbook: Railway reached nearly $100 million in annual recurring revenue with just 40 people. That's roughly $2.5 million of ARR per employee. In the age of AI, this kind of lean company will no longer be an outlier. So it's even more important for founders to study Railway and understand how Cooper built it.

This playbook breaks down 14 rules Cooper uses to run a lean company.

tldr: say no to 90% of ideas, do every job by hand before you hire for it, move one brick a week.

Inside you will find:

1. One decider, three in the room. Find out why Cooper never looked for a co-founder, and why the founder sits where the vote can go the other way.
2. Cap the free tier before the balance sheet does. Find out how Railway put a kill switch on a free tier that sold a dollar of infrastructure for about 20 cents.
3. Own the infrastructure your margins depend on. See how Railway left Google Cloud and had its first own servers running in about five months.
4. Solve it yourself, then check with users. Learn how Railway spends its few build-from-scratch bets, and where it borrows taste.
5. Say no to 90% of ideas with one line: why now? Find out why every idea needs a short abstract and a why-now line before Railway's quarterly planning week.
6. Lock the roadmap, then burn down lanes. See how 70% steak and 30% vegetables keep everyone shipping to the same plan.
7. Re-plan when the models jump. Find out why Cooper shelved an enterprise roadmap after one two-week winter break.
8. Do the job by hand, then hire for it. Find out why every Railway job description starts with work someone on the team already did.
9. Hire player-coaches and strong swimmers. Learn the one trait Railway screens every hire for, and the red flag it never ignores.
10. Never be stingy with talent. See why pay is reviewed every half and moves almost every quarter.
11. Test candidates on real work, and raise the bar with each model. Use Railway's bring-anything coding exercise.
12. Onboard in six weeks: trust, clarity, judgment. See the plan Cooper wrote in about 90 minutes on a Sunday.
13. Write values that check each other. Test your values against Railway's four, and set a 24-hour rule for incident posts.
14. Move one brick a week. Learn the weekly rhythm, including the 36 offline hours Cooper uses to find the biggest problem.

Rule 1. One decider, three in the room

Cooper's take: two is the worst number of co-founders. Two people disagree, a disagreement needs a tie-breaker, and nobody makes couples counselors for co-founders (Cooper's joke). The usual fix is to split the company by zone of excellence. That doesn't suit Cooper, who takes in input from every part of the business and finds consensus hard even internally.

So Cooper never went looking for a co-founder. Cooper likes a line from Carl Jung: do your work in earnest, and unknown friends will find you along the way. The founding engineer joined about a month after the company started, and the team grew from there. Three practices stand in for a second founder:

• One owner per decision. Railway uses directly responsible individuals (DRIs): one named person makes each call. (Note the similarity with the SPADE decision framework 😄)
• Rooms of three. Cooper sits in meetings with 3 people on purpose, so the vote can go against the founder. Plenty of times Cooper calls an idea terrible, the room disagrees, Railway does it, and Cooper turns out to be wrong.
• Fewer, better hires. The aim from day one was great people and fewer of them, so the company never reaches the point where hundreds of people can't decide.

Why it works: motion gives you information. A decision that gets made and corrected beats one that stalls. Our read: the DRI makes the decision fast, and the room of three makes it correctable.

📋 Founders: pick the decision that has been open longest in your company. Name one owner, invite two others to decide it with you, and set the date it gets made.