Founders · July 2026
Adeo Ressi's Epic Sprints
Most accelerators sell you a network and a demo day. Adeo Ressi built the biggest one on earth by selling the opposite thing: homework. A relentless weekly drumbeat of deliverables, graded by people who've actually built companies, with a real chance of getting cut if you don't ship.
The most demanding version of that drumbeat has a name inside the Founder Institute: Epic Sprints — custom assignments handed to you after a mentor review, aimed squarely at whatever is most likely to kill your company. I've been through the grind. Here's what the machine is, why it works, and how AI changes the math on every single sprint.
July 2026 · Alex Miller
The Man Who Turned Accountability Into an Industry
If you've raised on a SAFE, or seen a cap table with Class F founder shares, you've already touched Adeo Ressi's work. He was Elon Musk's housemate at Penn, sold a string of companies for close to $2 billion before he was thirty, then in 2007 built TheFunded — a site where founders anonymously reviewed their VCs, which was exactly as popular with VCs as you'd expect.
In 2009 he co-founded the Founder Institute with Jonathan Greechan. It is now the largest pre-seed startup program in the world — chapters in 250+ cities across dozens of countries, more than 7,000 portfolio companies, and north of $1.75B raised behind them. In 2021 he rolled the same philosophy up a level with Decile Group and VC Lab, which now helps launch a startling share of the world's new venture funds.
The through-line across all of it is one belief: founders don't need more inspiration, they need structured pressure and a deadline. Epic Sprints are that belief turned into a weekly assignment.
What an Epic Sprint Is
The Founder Institute core program runs roughly 14 weeks. Every week you owe a stack of deliverables — often 10 to 15 of them — done in working groups, defended in office hours, and either accepted or sent back. The standard curriculum walks the arc every founder has to walk: vision, customer development, revenue model, go-to-market, legal, product, hiring, funding.
Then there are the checkpoints. At the Mentor Idea Review (around week 5–7) and the Mentor Progress Review later, working founders sit in front of a panel that has built and exited real companies. They don't give you a grade and a pat. They give you Epic Sprints — extra, custom deliverables aimed at the specific hole in your business. Weak retention story? Your Epic Sprint is a cohort analysis and a re-engagement experiment. No proof anyone will pay? Your Epic Sprint is ten sales conversations and a signed letter of intent.
That's the part that makes it work. A generic curriculum teaches you the average lesson. An Epic Sprint is aimed at the one thing most likely to kill your company — and it comes with a date.
The quiet genius: the program is personalized by diagnosis, not by preference. You don't pick the sprint that sounds fun. A mentor who has seen a hundred companies die the way yours is about to picks the sprint you'd avoid. That's worth more than any deck of slides.
Why the Format Beats “Advice”
It's deliverables, not discussion
You don't leave the week with notes. You leave with a finished artifact — a synthesis, a landing page, ten booked calls. Things that exist can be graded. Opinions can't.
The deadline is real
Miss enough and you're out, refund window and all. Nothing focuses a founder like the possibility of being cut in front of their cohort.
It's aimed, not averaged
Epic Sprints target your specific weakness after a review. You do the work you were quietly hoping to skip, because a mentor named it out loud.
It compounds in public
You defend each deliverable to a group. Explaining your reasoning to peers every week is where the actual learning lives — the artifact is just the ticket in.
What AI Changes — and What It Doesn't
The Founder Institute itself has leaned hard into this — the current program runs an “Agentic Track” where the curriculum adapts to you with AI. That's the right instinct. When 10–15 deliverables a week used to mean 10–15 hours of grunt formatting, plenty of founders quietly cut corners. The bottleneck was never the thinking — it was the typing, the synthesis, the first draft of the deck.
That bottleneck is gone. An Epic Sprint that asks for a competitive teardown, a financial model, and a one-page synthesis is no longer a lost weekend. It's an evening with an AI partner doing the mechanical 80% while you spend your judgment on the 20% that's actually yours.
But — and this is the part I hammer in every workshop — AI does not do the sprint for you. The mentor panel can smell a generated answer from across the room. The deliverable was never the point; the decision inside it was. AI gets you to the decision faster, with more evidence, so you show up with a real point of view instead of a rushed one.
A Playbook for Nailing an Epic Sprint
Here's the loop I'd run on any custom deliverable, using an AI tool that can actually do work — Claude, or Claude Code if the sprint touches a repo or spreadsheet.
1. Turn the assignment into a contract
Before anything else, make the AI restate the deliverable in the exact format your mentor grades on, and list what it needs from you. Ambiguity is where sprints go to die.
You are my sprint partner for this week's Epic Sprint.
The deliverable: "Interview 10 target customers and synthesize the
top 3 problems worth building for."
Help me actually finish it, not talk about it:
1. Draft a 6-question interview script for <my ICP>.
2. Give me 10 outreach messages I can send today — LinkedIn + email.
3. As I paste raw notes back, tag each pain point and keep a
running tally so we can rank problems by frequency and intensity.
4. At the end, write the one-page synthesis in the format my
mentor wants: problem, evidence, who feels it, why now.
Ask me for anything you're missing before you start.2. Do the human part — the part that can't be faked
Send the ten outreach messages. Have the ten conversations. This is the irreducible core: no model can talk to your customers for you. What it can do is hand you the script and the target list so you're on calls by noon instead of still “preparing” at 5pm.
3. Let AI do the synthesis you'd have rushed
Paste the raw notes back and make it cluster, rank, and draft — while staying skeptical on your behalf. The blunt instruction is the important one: tell it to call out the “problems” that are really just features nobody asked for.
Here are my 10 raw interview transcripts (pasted below).
Cluster the pain points, rank them by how often they came up AND
how much emotion was behind them, and flag any quote that would
land in a pitch. Then draft the one-page synthesis. Be blunt about
which "problems" are actually just features nobody asked for.4. Pressure-test before the panel does
Ask the AI to role-play the toughest mentor on the panel and attack your synthesis. Every objection it finds tonight is one you answer calmly tomorrow instead of discovering live.
5. Ship early, bank the momentum
The whole design of the sprint is momentum. Finishing a day early doesn't mean you did too little — it means you can start the next sprint, or go deeper on the one thing the panel will actually care about. That compounding is the entire point of Adeo's machine.
The Real Lesson
Epic Sprints work because they replace “someday” with “by Thursday, in this format, defended out loud.” Adeo Ressi figured out a long time ago that the scarce resource in building a company isn't ideas or even money — it's the discipline to keep shipping the uncomfortable deliverable, week after week.
You don't need to be in a cohort to steal the format. Pick the one thing most likely to kill your project. Write it as a deliverable with a deadline and a grader. Then run the loop above with an AI partner so the mechanical work never becomes your excuse. That's an Epic Sprint of one — and it's the fastest way I know to find out whether what you're building is real.
Want to run your team's sprints with AI?
I run AI workshops that turn a week of deliverables into an afternoon — customer synthesis, models, decks, landing pages — with your team doing the judgment and AI doing the grunt work.
Book a 30-Minute Call