This post is my attempt to build something I can actually use when I start my next company. Think of it more like an operating manual.

(I’m a HUGE F1 fan, so gonna use some F1 analogies to explain my thought process here, please bear with me).

TLDR: The Framework

1. Choose your archetype. The Factory model (eg: Google; freedom to fail, outsized wins), the Pitcrew model (eg: Netflix; sports team, low margin for error), or the Amphibious model (eg: Open AI; hybrid). Pick based on what your company needs.

2. Lay the universal foundations. Psychological safety, dependability, clarity, meaning (the Patient Story), impact. Non-negotiable regardless of archetype.

3. Define your authentic layer. What’s actually true to you as a founder? What are you deliberately not optimizing for?

4. Propagate relentlessly. Hire in line with it, incentivise it, tell the story, make it visible. Culture decisions made early stick for years.

Where This Came From

Three conversations over the past few weeks sharpened my thinking on this.

First: Sebastien Dabdoub, a software engineer who’s worked at Google, Netflix, and OpenAI, in that order. The man basically speedran the culture spectrum.

Second: a talk at South Park Commons by a partner who’s been a founder and operator across eight companies.

Third: Siddharth Seth, founder of Pavo AI, who’s living this problem right now in real time.

Phase 1: What Kind of Company Are You?

Most founders never explicitly choose a culture. They just drift into whatever the first five employees vibe with, and then keep fixing when things feel off at 20+ people.

Sebastien broke it down into three archetypes, and I haven’t found a cleaner way to think about it:

The Factory model (eg. Google): fail until you don’t. You can take wild bets, fail nine times, and nobody panics. Because when the tenth bet works, you get Waymo or DeepMind. It’s how F1 teams come up with breakthroughs for their cars back at Brackley or Maranello. This generally works if you have a cash cow (Google has Search) bankrolling the experimentation. Without that, you’re just failing with no safety net.

Google also cracked something nobody else has replicated: keeping people who could easily go start their own companies. They let engineers spend one day per week working on whatever they wanted, on company time. If there was some outrageous success, they would compensate them proportionally. It worked. People who were rich enough to retire and talented enough to start anything stayed for years.

The Pitcrew model (eg. Netflix): perform or get cut. Sports team culture. You’re on the roster because you’re excellent. The moment you stop being excellent, you’re off. This produces insanely efficient teams. But Netflix will never build a Waymo. The culture can’t support high-failure-rate innovation with long time horizons. You’re trading ceiling for floor.

(Worth noting: Netflix isn’t innovation-free. They built and open-sourced the streaming framework the entire industry uses. But all their innovation happens within a tight strategic frame. Every word in their strategy document is stress-tested. If it’s not in the strategy, it doesn’t happen. Pitcrews in F1 will innovate to the world’s end to save every 0.1 second possible. But they can’t fundamentally alter the dynamics of it such that you gain an insurmountable advantage over competitor cars.)

The Amphibious model (eg. Open AI): lab + product engine. The research lab part of Open AI runs like Google: long timelines, high failure tolerance, because the wins are civilization-level. Their product team runs like Netflix: ship, optimize, grow, no excuses.

Two operating systems, one company. Someone has to manage the seam.

I saw this playing out at Pavo AI. Siddharth has a PhD-led science team and a product/engineering team under the same roof. The scientists work in deep focused bursts; their “deliverables” are breakthroughs that might take months. You can’t sprint-plan your way to a novel insight. Meanwhile, the product team needs to ship fast and continuously. Completely different rhythms. Siddharth is the bridge between both, and his job is making sure neither side resents the other for operating differently.

If you’re building an AI company in 2026, there’s a decent chance you’ll need to solve this exact problem.

Important: choose before you hire. If you don’t, the culture defaults to whatever emerges accidentally. And accidental culture is how you end up with a 15-person company that feels like a group project where half the team didn’t read the brief.

Phase 2: The Universal Foundations

Whatever archetype you pick, if you’re building an innovation company (you probably are), there’s a bedrock layer that applies to every innovation company.

Google’s Project Aristotle studied 180 teams and found five things that predicted success:

1. Psychological Safety

Can the most junior person in the room tell the most senior person they’re wrong? Not a “well, maybe we could perhaps consider...” but a “I have data, you’re wrong, here’s what I’d do instead.”

People always bring up the supposed counterexamples. Steve Jobs called people’s work garbage. Musk is... Musk. Jensen Huang talks about torturing people to greatness.

But I’d argue that they’re the strongest evidence for the point. Each one of these leaders care about radical truth. The best idea wins regardless of who said it. A junior engineer at SpaceX who finds a better structural solution gets heard. People in these companies compete to surface the right answer. That IS psychological safety. It’s just wrapped in intensity.

The culture fails when people keep quiet, especially because speaking up carries political risk. It kills companies, and it’s fairly common tbh. It starts with you. Admit when you’re wrong, out loud. Reward the person who pushes back on you with better data.

2. Dependability

If you say you’ll do something, do it. If you can’t, say so immediately. Not at the deadline, or the day after. Because the cost of getting this wrong is cognitive load (apart from obviously missing the deliverable deadline). When I hand something off and know you’ve got it, my brain lets go. When I’m not sure, I’m carrying that weight on top of my own work. Multiply that across 10 people and you’ve got a company running at half speed while looking busy.

3. Clarity

Everyone needs to know what direction they’re paddling. Not once. Constantly.

Founders always underestimate this because they made the decision and have been marinating in it for weeks (I’ve personally made this mistake). It feels obvious. Meanwhile, if you ask your 12-person team what the top two priorities are, you’ll get eight different answers. You think you’re a dragon boat pulling in formation when in reality, you’re barely even afloat.

Over-communicate. You will feel like it’s too much. It’s not. It’s barely enough.

4. Meaning (the Patient Story)

Everyone on your team could go get a comfortable job tomorrow. They’re with you because of the mission.

Best example I heard: a cancer diagnostics company that opened every all-hands with a patient story. Not revenue. Not metrics. A patient who was alive because of their work. The chief medical officer would show scans. Before <> after. Tumor <> then no tumor. People still get chills years later.

You might think your B2B SaaS doesn’t have patient stories. Find the human story anyway. If you’re saving someone three hours a day, you’re giving them evenings with their kids. That counts. Lean into it. Humanise your company! (even more important in this post-modern sci-fi AI slop era).

5. Impact

People need to feel and understand how their specific contribution matters to the above. Not just “great mission” in the abstract, but “the thing I built last week directly caused this.” Make that connection visible. Don’t let great work disappear into the void.

Phase 3: The Part That’s Authentically Just You

Universal foundations are table stakes. On top of them, you build something authentic to who you actually are.

If your culture isn’t a genuine reflection of how you operate, it won’t survive the first hard quarter. People detect inauthenticity fast.

Zuckerberg built “move fast and break things” because that’s who he is. Engineers checking in code by day two. That would be insane at a medical device company. But it built Facebook.

Tesla/SpaceX is hardcore intensity. Authentic to Elon. Not replicable by people who aren’t Elon, and that’s fine.

Judith Faulkner at Epic Systems put “Do not go public. Do not be acquired. Be frugal.” on the wall. Very Midwestern. Very deliberate. Epic is one of the most valuable private companies on earth today.

GitLab went fully remote because everyone was remote when they started. They then leaned into it and turned it into an advantage.

Some choices are mutually exclusive. “Move fast and break things” can’t coexist with “zero defects.” If you want wild experimentation, you can’t also demand nothing ever breaks. The sign that you’ve made a real cultural choice is that you can articulate what you’re giving up.

Start with yourself. What do you value? How do you work best? What drives you crazy? Your personal values are the seed. Not all of them should make it into the company (some you should hire against), but the core should feel natural. Because if it doesn’t fit you, you’ll either exhaust yourself pretending, or the mask slips and trust goes with it.

Here’s an exercise that I’ll do the next time I start a company (I’d recommend you to do this too): Write down what values / working styles / culture ethos are important to you as a person? (Forget about the company for a second). Crystallise them. The company culture & values should be an extension of this.

Phase 4: Making It Stick

Defining culture is maybe 20% of the work. The rest is making actually sticking with it.

Hire explicitly. If your culture is intense, say so in the interview. “We’re here until midnight fixing things and we love it.” Let people who don’t want that opt out before they’re on your team. The rule: if you can’t say a cultural criterion out loud in an interview, it shouldn’t be a criterion. (This also protects you from disguising “people who look like me” as “culture fit.”)

One example that stuck with me: a tech company in Israel hired primarily from the local ultra-orthodox community. Engineers with six or seven kids, not working 9-9-6. But brilliant and hard-working during their hours, and because no other tech company was hiring them, the company had access to a talent pool everyone else was ignoring. There are many ways to build a company.

Reward the right behaviours. Every time someone does what your culture is supposed to produce, call it out. Publicly. Immediately. A Google t-shirt from 2003 is worth more than a designer suit, because you can’t buy one; you had to be there. Mission patches, public acknowledgment, a quick “great call” in standup costs nothing, but matters enormously.

Tell the founding story. Again and again. The origin story encodes your values without feeling like a lecture. Tweak the emphasis to highlight what you want to propagate in terms of your culture & values. If you’ve been at any well-run company, you’ve heard the founding story multiple times. It may feel a bit self-indulgent, I know, but it’s a cultural transmission mechanism doing its job.

Make what matters visible. Qualcomm has a wall of patents in their San Diego HQ. Most important patents get the biggest spots. Walk in that door and you understand in five seconds what the company is about. What’s visible in your Slack, your all-hands, your onboarding? That’s your actual culture. Nobody reads the values doc anyway.

Hire against your own weaknesses. The company will reflect you, good and bad. If you tend to micromanage, or avoid conflict, or make impulsive calls when excited, hire leaders who buffer the team from your worst tendencies. One of the most honest descriptions I heard of a co-founder relationship: “A big part of my job was protecting everyone from the his (the CEO’s) randomizations.” Know what you’re bad at. Hire against it. Give yourself grace; being a founder is one of the hardest jobs in the world!

One More Thing

Gridware introduced “every other Monday off” when they were five people. Today, they are a 140 member team and have gone through multiple growth stages, but this policy has still stuck. Became a core recruiting advantage.

Culture decisions that are made early deliberately stick around way longer than you’d expect.

At some point you’re not in the room for half the conversations. Then most of them. Then almost all of them. What happens in those rooms when you are not present, is your culture. It’s either the operating system you designed, or it’s whatever randomly showed up when you weren’t paying attention.

Make it deliberate.
Thank you for coming to my TED talk.

On the Market / Problem Statement

1. AI isn’t a category, it’s the substrate. 60% of the batch is AI-native. Another 26% is AI-enabled. Only 14% has no AI. The question isn’t “are you using AI?” It’s “what does your AI do that a foundation model can’t do out of the box?”

2. Replace, don’t assist. The defining theme is “AI employees,” not copilots, not assistants. The pitch is always “we replace [expensive human role] end-to-end,” priced as a fraction of that person’s salary. Copilots assist. Agents act. The industry moved on.

3. Find the “Claude Code” for your domain. Every profession has structured outputs AI can now generate: contracts, CAD files, financial models, surgical plans, specifications. Find a profession where practitioners cost $100-500+/hr, tools are 10-30 years old, and there’s a clear verification step. Wide-open domains: tax prep, civil engineering, management consulting, clinical trials, patent drafting, music production.

4. Consider the service model. ~20% of the batch is building AI-native service firms (law, recruiting, accounting, insurance) that charge for outcomes with software margins. They show the fastest revenue in the batch. The playbook: start as a service → get revenue and data → ship the automation → graduate to a platform.

5. B2B dominates. AI agents replacing B2B knowledge workers. 87% B2B. Only 14 consumer-facing companies (~7%). The current AI capability unlock maps perfectly to business workflows. This is a good trade, but the legendary companies from this batch will likely be the outliers: the uranium explorer, the moon hotel, the robot cowboys, the parasite drug company.

6. Build a data flywheel. Every customer interaction should make your product better. LegalOS trained on 12K visa petitions → 100% approval rate. Perfectly improves with every hire. Without a data flywheel, you’re a wrapper.

7. Don’t build a general-purpose AI wrapper. “AI for everything” loses to “AI that replaces one specific $80K/yr role.” Go deep into an unsexy industry. The best opportunities are in industries you’d never pitch at a cocktail party.

8. The absence of consumer is the opportunity signal. Zero education companies. Zero consumer social. Zero mental health/wellness. Zero govtech. The least-funded categories historically produce the largest outlier returns. The founder who cracks AI-native entertainment, social, or education has the category to themselves.

9. Hardware is back. 18% of the batch has hardware components (robotics, drones, wearables, space tech). That’s a notable jump from recent batches. The SpaceX/Tesla alumni building physical products are among the most differentiated companies in the batch.

On Distribution

10. Distribution is a prerequisite, not an afterthought. 60% of the top 15 companies by growth acquired customers through founder network or YC network. If your first 20 customers require “figuring out distribution,” you picked the wrong market.

11. Your previous employer is your first market. The dominant GTM motion (~35% of B2B): founders who spent years in an industry, left, and sold back into their network. Their Rolodex IS the distribution channel.

12. The PE roll-up channel is massively underexploited. Ressl AI and Robby independently discovered that PE-backed roll-ups are desperate for margin improvement tools. One PE deal = 50-200 locations.

13. Choose markets where you already have a distribution network. The companies struggling with GTM are almost always the ones that built a product first and then asked “how do we sell this?” The winners asked “who do I already have access to, and what do they desperately need?”

On Teams

14. Founder-market fit is the single strongest predictor of speed to revenue. The founders who literally did the job they’re now automating close deals in days. Everyone else takes months. Proximitty ($700K ARR in <3 weeks): CEO was a McKinsey bank risk advisor. Corvera ($33K MRR in 4 weeks): CEO ran a CPG brand.

15. Your cofounder relationship is your moat. 46% of the batch is 2-person teams. The strongest teams worked together for years: former colleagues, classmates, siblings, repeat cofounders. If you haven’t shipped something with your cofounder, you haven’t validated the most important part of your startup.

16. Domain expertise beats pedigree. The most compelling founders lived the problem: the dentist building surgical AI, the aircraft maintenance head building mechanic tools, the lobbyists building policy AI. “Ex-FAANG” is table stakes, not a differentiator.

On Pitching

17. The crazy closer matters. When 199 companies pitch in one day, you need to be the one they talk about at drinks. “The first AI Oscar will be made on Martini.” “You can book a hotel on the Moon for 2032.” Make your vision specific, falsifiable, and quotable.

On What to Avoid

18. Avoid undifferentiated agent infrastructure. 8-10 companies building agent monitoring/testing/compression. Foundation model providers will build these natively. If “[Existing DevOps tool] but for AI agents” describes you, that’s the danger zone.

19. Avoid AI-native services without a data moat. Fastest revenue but lowest defensibility. Core tech replicable in weeks. Traditional firms will adopt AI in 12-18 months. Without proprietary data or embedded distribution, the moat is thin.

20. Avoid commodity workflow wrappers. AI for a single well-defined task where GPT-5 might do the same thing natively in 6 months.

The Scene

199 pitches. Something about freshly baked startups coming out of the YC oven that just smells so good. Exciting, high-energy, never a dull moment.

Some moments that stuck:

• A startup pitching the first hotel on the Moon, with White House invitations and $500M in LOIs

• Robot cowboys herding cattle with autonomous drones

• An AI presentation company generating its pitch deck live during the demo

• One company casually zooming into Tehran, Iran while demo-ing satellite imagery (the room went quiet)

• Martini’s founder closing with “The first Oscar for an AI-made movie will be won by Martini!”, the kind of line that makes investors either roll their eyes or reach for their checkbook

The hardware demo area was buzzing: robots, drones, microscopes with lifesciences proteins, radars mounted on cars. Real, physical things you could touch. This wasn’t just a batch of SaaS dashboards.

After 199 pitches, you stop hearing individual companies and start seeing patterns. Here’s what I found.

The Macro Numbers

Total companies: 199

Business Model:

• B2B: 174 (87%)

• B2C: 14 (7%)

• B2B2C: 11 (6%)

Product Type:

• Pure software: 163 (82%)

• Hardware + software: 24 (12%)

• Pure hardware: 12 (6%)

AI Classification:

• AI-native (AI IS the product): 120 (60%)

• AI-enabled (existing workflow + AI): 52 (26%)

• Non-AI: 27 (14%)

Traction:

• Estimated median ARR: ~$50-100K

• Estimated median growth: ~30-50% MoM

• Companies >$1M ARR: ~5%

• Pre-revenue: ~50%

Top industries: B2B software (59%), Industrials (15%), Healthcare (10%), Fintech (8%), Consumer (4%).

Only 14 companies are consumer-facing, and YC officially classifies just 7 as “Consumer.” The rest are consumer products wearing enterprise labels, tucked under B2B, Healthcare, or Fintech.

The Ten Themes

1. AI Agents Replacing Entire Job Functions

The defining theme. Not copilots, full replacement.

• Beacon Health replaces admin staff doing prior authorizations

• Perfectly replaces recruiters end-to-end

• Lance replaces hotel front desk across 50+ Marriott/Hilton/Hyatt properties

• Mendral (Docker’s co-founders) replaces the DevOps engineer

• Canary replaces QA

The “copilot” framing dropped from ~4% of pitches in early 2025 to 1% in W26.

2. “Claude Code for X”

Claude Code and Cursor proved agentic AI works for code. W26 founders are applying the same paradigm to every profession with structured outputs:

• REV1 for mechanical engineers (3D→2D drawings)

• Avoice for architects (specifications, documentation)

• Synthetic Sciences for scientific research

• Maywood for investment bankers

• Alt-X for RE underwriting (works directly in Excel)

• Cardboard for video editing

• Mango Medical generates surgical plans in minutes instead of days

3. AI-Native Professional Services (”Service Business, Software Economics”)

Not building tools for incumbents, building AI firms that compete with them:

• Four AI law firms (Arcline, General Legal, Vector Legal, LegalOS)

• AI recruiting agency (Perfectly)

• AI accounting (Balance)

• AI insurance brokerage (Panta)

• AI policy consulting (Fed10, founded by three ex-lobbyists)

Panta says it explicitly: “A service business with software economics.” Charge for outcomes, operate with software margins, because AI does 80% and humans handle 20%. Arcline has 50+ startup clients. LegalOS has 100% visa approval rate.

The bear case: humans-in-the-loop cap margins at 60-80%. Liability is real. The moat question: if the core tech is “LLM + domain prompts + human review,” what stops replication? The emerging answer: start as service → ship automation → graduate to platform. The service is the wedge; the software is the moat.

4. Infrastructure for the Agentic Era

Every tech stack layer is being rebuilt for agents:

• Agentic Fabriq = “Okta for Agents”

• Sponge (three ex-Stripe crypto leads) = financial infrastructure for agents

• Moda/Sentrial = Datadog for agent reliability

• Salus = runtime guardrails

• 21st (1.4M developers) = React components for AI-first UIs

• Zatanna turns pre-LLM SaaS into databases agents can query

The risk: foundation model providers build these natively. The ~30% competitive overlap in this layer confirms it’s crowded.

5. Vertical AI for “Unsexy” Industries

The biggest ROI is in industries tech has ignored:

• Zymbly automates aircraft maintenance paperwork (45 min docs per 5 min repair)

• GrazeMate builds robot cowboys, autonomous drones herding cattle. When they pitched, you couldn’t help but grin. It sounds absurd until you learn the founder grew up on a 6,000-head cattle station.

• OctaPulse does CV for fish farming

• Squid tackles power grid planning ($760B annual inefficiency, still on spreadsheets)

These founders went deep. Scout Out’s founder is fourth-generation construction. LegalOS’s co-founders grew up in their family’s immigration law firm (10,000+ hours each since age 12). Zymbly’s co-founder was Head of Aircraft Maintenance at Virgin Atlantic. The best opportunities are in the industries you’d never pitch at a cocktail party.

6. Physical AI / Robotics Renaissance

18% of the batch has hardware components:

• Remy AI and Servo7 build warehouse robots learning from human demonstrations (80% of warehouses have zero automation)

• Origami Robotics builds robot hands

• RoboDock went viral deploying their MVP in 60 days, landed a $100K Waymo contract

• Fort (three ex-Tesla engineers) tracks strength training, something Whoop/Oura still can’t do

• Pocket shipped 30K+ units, $27M annualized revenue

The hardware demo area was the most energizing part of the day.

7. Defense & National Security

• Milliray (three Oxford/St Andrews PhDs) builds drone detection radar for NATO ($470K in batch sales)

• Seeing Systems builds AI strike drones for UK Royal Marine Commandos

• DAIVIN! builds tankless dive gear for U.S. Special Ops

Defense budgets are large, contracts long, credibility transfers to commercial.

8. Data as the Moat

When everyone has the same foundation models, proprietary data is the primary defensibility:

• Shofo: world’s largest indexed video library

• Human Archive: dropped out of Stanford/Berkeley, moved to Asia, collecting data from thousands of homes for humanoid robots

• LegalOS: 12K successful visa petitions → 100% approval rate

The pattern: every customer interaction makes the product better. Without a data flywheel, you’re a wrapper.

9. Hard Tech & Space

The most audacious pitches. GRU Space is building the first hotel on the Moon by 2032. When they pitched, the room recalibrated: half thought they were crazy, half thought they might do it. $500M in LOIs, White House invitations, 1B+ views. Beyond Reach Labs builds football-field solar arrays for orbit (500x power demand increase by 2030). Terranox uses AI to discover uranium deposits (single discovery = $200-700M).

Ditto Biosciences might be the most creative thesis: parasites evolved proteins that control the human immune system over millions of years. Ditto uses AI to identify them and engineer autoimmune therapies. Evolution already solved the problem, they’re just reading its answers.

10. AI-Native Research & Science

• Talking Computers deploys fleets of AI scientists ($1M+ ARR)

• Aemon (twin brothers, published at ICLR/EMNLP before age 20) set a world record on an NP-hard math problem with <$10 compute, beating Google DeepMind

• Ndea, co-founded by Zapier’s Mike Knoop and Keras creator François Chollet, is explicitly building AGI that can innovate

The Founders: Patterns From 429 People

Demographics:

• ~60% immigrant/international

• 86% male, 14% female

• Top schools: Berkeley (~45), Stanford (~35), MIT (~20), Waterloo (~15)

• 55% studied CS; 45% didn’t

Backgrounds:

• ~30% ex-FAANG

• ~25% had a previous startup

• ~12% ex-finance/trading (Citadel, Jane Street, Jump)

• ~12 founders from SpaceX alone, overwhelmingly building hardware and aerospace

Teams:

• 46% are 2-person teams, 15% solo

• Most common archetype: two technical cofounders with different specializations (~35%), not the classic “hacker + hustler”

• 19% of companies have at least one PhD founder

• How they met: ~35% university classmates, ~25% former colleagues, ~15% repeat cofounders, ~10% family/siblings

Domain experts who became founders are the most compelling stories: Adrian Kilian (dentist → surgical AI at Mango Medical), Robbie Bourke (25 years in aviation → Zymbly), Pamir Ehsas (outside counsel to OpenAI → Arcline), Conor Jones (years inside National Grid → Squid).

Some observations:

• Deep domain expertise + a technical co-founder who can build = the strongest companies in the batch

• The most successful-looking teams either built and sold a company together before, or worked side-by-side at the same company on the same problem they’re now solving

• 31% of companies have at least one PhD or researcher founder, primarily concentrated in healthcare/biotech, hard tech, and AI infrastructure

How They Found Their Markets

B2B (88% of batch)

• “I Lived This Pain” (~40%): The strongest pattern. End Close’s founder spent 6 years at Modern Treasury processing $1T+ in payments. Squid’s founders spent years inside National Grid. They didn’t need customer discovery, they were the customer.

• “I Built the Platform This Replaces” (~20%): Docker’s co-founders building Mendral. TikTok’s ML scientists building Perfectly. They know the architecture intimately and see where AI creates a step change.

• “The 50-Conversation Sprint” (~15%): Systematic discovery. Ritivel had 50+ pharma conversations before writing code. Ressl AI started as consulting, found trades had the most glue work.

• “Infrastructure Prophecy” (~15%): Thesis-driven. “If agents exist, they need auth” → Agentic Fabriq. Risk: building for a future 2-3 years away.

• “Research → Commercialization” (~10%): CellType (Yale professor + DeepMind). Valgo’s co-founder literally wrote the textbook on safety-critical systems.

B2C (7% of batch)

• “I Am the User” (~50%): Fort founders are lifters frustrated by wearables. Doomersion’s founder doomscrolls and studies languages, combined them.

• “Format Shift” (~25%): Existing behavior + new medium. Pax Historia: love of strategy games + AI alternate history.

• “Hardware Wedge” (~25%): Physical product creates data loop software can’t replicate.

The meta-lesson: Not a single successful W26 company was born from a hackathon or a “what if we used AI for...” brainstorm. Every one traces to deep personal experience or obsessive customer discovery.

How They Found Distribution

The data is clear: founder network is the #1 mechanism for the fastest-growing B2B companies. 60% of the top 15 by growth rate acquired first customers through founder network or YC network.

B2B patterns:

• “Sell to former employer’s peers” (~35%): Fed10’s three ex-lobbyists, their Rolodex IS distribution

• “YC as launch pad” (~25%): Cardinal runs outbound for 40+ YC companies, Palus Finance signed 33 in weeks

• “Open source” (~10%): 21st has 1.4M developers, only works for infrastructure

• “PE roll-up channel” (~8%): One deal = 50-200 locations

• “Systematic outbound” (~15%): Finite buyer lists with quantifiable pain

• “Wedge product” (~7%): Land narrow, expand everywhere

B2C: The product IS the distribution. Doomersion got 15K downloads in 2 weeks, zero paid marketing. Pax Historia built tens of thousands of DAUs organically. Hardware founders bet that physical presence generates word-of-mouth.

The biggest takeaway: The companies struggling with GTM are almost always the ones that built a product first and then asked “how do we sell this?” The winners asked “who do I already have access to, and what do they desperately need?”, then built that.

Anatomy of a Great Pitch

Seven components that separate pitches that stuck from ones that blurred:

1. Hook. Three archetypes work:

• The Shocking Stat: “It takes 500,000 days to bring a drug to market. We want to make it 5” (Rhizome AI)

• The Reframe: “Every file you’ve ever uploaded uses a protocol from 1974” (Byteport)

• “I Am the Problem”: “I spent 6 years building reconciliation at Modern Treasury processing a trillion dollars” (End Close)

2. Problem (specific, not generic). “Technicians spend half their time on paperwork” (Zymbly) beats “We automate back-office workflows.”

3. Team (one-line credibility bombs). “Andrea wrote Docker’s first lines of code” (Mendral). “Our team invented the MPIC standard securing every HTTPS connection on the internet” (Crosslayer Labs).

4. Market (inevitable, not just large). “Satellite power demand: 500x increase by 2030” (Beyond Reach Labs). The strongest market pitches explain why now and why this is inevitable, not just how big the TAM is.

5. Traction (velocity > absolute number). “$0 to $33K MRR in 4 weeks” (Corvera) beats “$100K ARR” with no timeframe.

6. Unique insight. “Parasites evolved proteins that control the human immune system. We read their answers” (Ditto Bio). “Insurers can’t price autonomous systems because historical claims data doesn’t exist” (Valgo).

7. Crazy closer. “The first AI Oscar will be made on Martini.” “Book a hotel on the Moon for 2032” (GRU Space). The pitches that blurred: generic “AI for [industry],” team credentials without connection to the problem, and (critically) no crazy closer.

Competitive Overlap: YC’s Multiple Bets

~30% of companies have a direct competitor in-batch. Only ~5% face truly high overlap.

• High overlap: LLM context compression (Token Company vs. Compresr), med-legal docs (Wayco vs. Docura Health), robotics data (Human Archive vs. Asimov)

• Medium: Startup legal (Arcline vs. General Legal vs. Vector Legal), AI SRE (IncidentFox vs. Sonarly), agent monitoring (Sentrial vs. Moda), prior auth (Ruma Care vs. ClaimGlide vs. Beacon Health)

What it tells you: YC bets on markets, not companies. Three startup law firms = the market is real and large enough for multiple winners. Two identical-looking companies on Demo Day will look completely different by Series A. The most differentiated companies have zero overlap: Terranox, Zymbly, GrazeMate, Ditto Bio. In each case, the founder’s domain expertise IS the moat.

What’s Conspicuously Absent

• Zero education companies

• Zero govtech

• Zero consumer social

• Zero mental health/wellness

• Near-zero marketplaces

• Near-zero pure crypto (blockchain used as plumbing, never as product thesis)

• Consumer at a historic low (14 companies total, just 7 officially classified)

Industrials went from 3.6% of W24 to 14.1% of W26, a 4x jump. The “atoms vs. bits” shift is real within YC.

The contrarian read: W26’s composition is a snapshot of what’s fundable right now, not what will be valuable in 10 years. The legendary companies missing from this batch are the consumer and social founders who will arrive in 2-3 batches, once AI capabilities catch up to their ambition.

What’s Likely to Fail

• Undifferentiated agent infrastructure. 8-10 companies in agent monitoring/testing/compression. Foundation model providers will build these natively. Enterprise buyers default to existing vendors.

• AI-native services without a data moat. Fastest revenue, lowest defensibility. Core tech replicable in weeks. Traditional firms adopt AI in 12-18 months.

• Solo technical founder in relationship-sold markets. Construction, insurance, freight: if nobody can walk into a job site and speak the language, it stalls.

• “AI for [Industry]” without domain depth. The tell: description leads with “We use advanced LLM agents to...” rather than the customer’s specific pain.

• Pre-revenue deep tech with long horizons. Not wrong conceptually, but the failure mode is running out of runway.

• Commodity workflow wrappers. Single-task AI where GPT-5 might do the same thing natively in 6 months.

The Fastest Companies Share Five Traits

1. Sold the outcome, not the tool

2. Founder had customer relationships before the product existed

3. Charged from Day 1: no free tier, no pilot purgatory

4. Customer was desperate, not curious (Proximitty: banks with $2B+ delinquent loans; Ruma Care: clinics denied $150K reimbursements)

5. MVP was embarrassingly simple: they describe outcomes, not architectures

The gap between “launched and learning” and “building and hoping” is where most of the mortality in this batch will occur.

Exciting times ahead! Never been a better time to build.

Written on March 25, 2026, days after YC W26 Demo Day.

Three completely different meetings this week. Three completely different worlds: space cannons, self-driving cars, and personalized medicine. Each one left me with something I don’t want to forget.

TLDR of founder lessons:

1. You cannot beat a giant at their own game. Take asymmetric, contrarian bets.

2. If the physics is theoretically possible, every failure is a skill issue.

3. First-principles people don’t rely on others’ opinions. They read, build, experiment, then form conviction & back it to the hilt.

4. Founder-problem fit beats market analysis. The best founders lived the problem.

5. De-risk the co-founder question early. Build with someone you’ve already built with.

6. GTM has to start by picking a wedge, not selling to the whole market. Also, distribute via believers first.

7. San Francisco rewards showing up. Every conversation is a potential door.

1. The Space Cannon Guy

I was working at a WeWork when I noticed a guy having lunch alone. I walked up, made small talk, asked him what he does.

His name is Mike Grace. He runs a company called LongShot Space.

Mike is an ex-Army veteran, deployed twice in Afghanistan. He built his last company and sold it to the Department of Defence for millions. His most recent project before LongShot? Helping Ukraine build missile attack systems to fight Russia.

That’s the resume of the guy casually eating lunch at a WeWork.

He invited me to a cigar lounge that evening. Over fine whisky and cigars (he taught me how to pick, cut, and light one), he explained what LongShot does.

The concept: There are five known methods to send a payload into orbit. Rockets are just one of them, and a deeply inefficient one. Only about 5-7% of a rocket’s total weight can be payload. The rest is fuel: and most of that fuel is just there to lift the other fuel. This is the tyranny of the rocket equation, and it’s what makes rockets inherently expensive.

LongShot is building a space cannon. Literally. A massive pneumatic gun where the “bullet” is a satellite or payload. The projectile accelerates through a barrel using compressed hydrogen gas, injected from the sides in stages, until it reaches orbital speeds. In this system, up to 50% of total weight can be payload: making launches potentially 50x to 100x cheaper than rockets.

Thanks to Andre, who handles business development for LongShot, for showing me around their prototype facility at Alameda: a former U.S. Navy cannon testing facility just outside San Francisco. The prototype is a steel tube, roughly 100 meters long with a six-foot diameter. I didn’t see it fire, but just seeing the physical setup was wild.

The strategic bet: SpaceX is reportedly gearing up for an IPO around mid-2026, potentially the world’s largest, which could make Musk a trillionaire. SpaceX would effectively become the monopoly on sending satellites to orbit. It nearly is already.

Mike’s positioning is simple: when SpaceX IPOs, he wants to be the challenger. The guy investors look at and say, “Okay, there’s someone else.” But you cannot challenge Musk by building cheaper rockets. That’s his game, and he’s been at it for over two decades. You challenge him with a fundamentally different technology. An asymmetric bet.

The “why now” question: If the physics of space cannons has been known since World War II (the German V-3 cannon used the same multi-injection principle), why is this happening now?

The answer isn’t physics. It’s the market. SpaceX created an explosion in satellite demand (Starlink alone has 6,000+ satellites). The U.S. Air Force is investing in alternative launch methods. Sam Altman, Tim Draper, and Space Fund are backing LongShot. In the 1990s, there was no market demand, no investor appetite, and no institutional support for “a giant space cannon.” Now there is.

The irony: Musk built the very market conditions that make his own disruption possible.

The Lesson: You Cannot Beat a Giant at Their Own Game

If you’re going up against a dominant player, you don’t compete on the same axis. You find a completely different approach where you have an unfair advantage and they have a structural disadvantage. SpaceX is optimized for rockets: their entire infrastructure, team, IP, and capital structure is built around rockets. A space cannon isn’t a marginal improvement: it’s a category reset. That’s the only kind of bet that has a shot against a monopoly.

Asymmetric doesn’t mean small. It means different. Mike isn’t building a hobby project. He’s building a 10-kilometer cannon designed to shoot things at Mach 29. The ambition is enormous. The approach is just orthogonal.

2. The Tesla Autopilot Guy

Through Viram (CEO of Vested), I got connected with Dhaval Shroff. Dhaval is an Autopilot AI engineer at Tesla: not just any engineer. He’s the guy who, in December 2022, pitched the idea of an end-to-end neural network planner to Elon Musk: essentially “ChatGPT, but for cars.” That conversation is documented in Walter Isaacson’s biography of Musk. Dhaval is described as a trusted aide to Musk.

His background: Indian. Undergrad in Mumbai, Master’s in Robotics at CMU. Joined Tesla Autopilot as an intern in June 2014. Converted to full-time. Moved from driver assistance, to Autopilot R&D, to Autopilot AI. About 12 years now. One company. One bet.

Why he stayed: I asked him: how is it that you’ve been at Tesla for over a decade and never left? Plenty of self-driving companies would have thrown money at him.

His answer: conviction in a contrarian bet.

The context: twelve years ago, every serious self-driving project incl. Waymo, and everyone following Waymo’s lead was building on LIDAR and radar systems. Tesla was the only company that said we’re going pure vision plus neural networks. No LIDAR. No radar. Just cameras (the “eyes”) feeding data to a neural network (the “brain”): mimicking how humans actually drive.

Musk’s reasoning: humans drive using their eyes and their brain. If we want cars to truly drive like humans, replicate that loop: optical input, neural processing, action output. Not laser beams bouncing off objects.

This was wildly contrarian at the time. Almost every expert thought Tesla was wrong. Dhaval had options to join other self-driving companies, but they were all going the Waymo way. He figured if he was going to take a bet, he’d rather take the contrarian one.

On Andrej Karpathy: Karpathy joined Tesla as head of AI in 2017 and was another reason Dhaval stayed.

What made Karpathy special: he is incredibly hands-on. He doesn’t rely on other people’s opinions or someone else’s research summaries. He reads the white papers himself. Builds his own logic on top of them. Gets his hands dirty and experiments personally. Only then does he form an opinion. But once he does, his conviction is absolute, because it’s grounded in first-principles work he did himself.

Karpathy was also deeply interested in what every team member was building. He’d sit with engineers, including on weekends. That level of involvement and excitement was contagious. He wasn’t a figurehead: he was in the trenches.

On Elon Musk: Dhaval’s take is that the guy is ruthless. His operating principle is deceptively simple:

1. Determine what is possible according to the laws of physics.

2. If the physics says it’s theoretically possible, then any failure to achieve it is an execution problem: a skill issue.

When Musk decided to go all-in on the vision-only bet, many team members told him it probably wasn’t possible. His response: “Is the physics theoretically sound?” Yes. “Then it’s a skill issue. Go find better people.”

He recruits the best people from the best universities and companies. He finds capital from the best sources. And he goes absolutely relentless on the mission.

On Musk’s product philosophy: Every product Musk builds has to have civilizational-scale impact. He doesn’t do niche. Tesla had to be affordable enough for every person on Earth to eventually own one. Same logic across all his companies (space migration for everyone etc). If the technology can’t serve every human on the planet, it’s not worth his time. (Which is why he doesn’t do Louis Vuitton type products).

What people get wrong about Tesla: It always comes back to the contrarian bet. People underestimated the vision-only approach for years.

Once you have conviction in a bet like that, you back it to the hilt. All in.

The Lessons

Contrarian conviction is the real moat. Dhaval could have joined any self-driving company. He stayed at Tesla because he believed in the contrarian approach. You don’t beat the incumbent by doing what they do slightly better. You beat them by doing something fundamentally different that they can’t easily switch to.

First-principles people are rare and magnetic. Most senior leaders in tech operate on summarized information: someone else’s research, someone else’s analysis, someone else’s opinion. Karpathy reads the papers, builds the prototypes, runs the experiments, and then forms an opinion. That’s why people follow him. As a founder, this is the standard I want to hold myself to.

“Is the physics possible?” is the only question that matters. Musk’s framework is brutal but clarifying. If something is theoretically possible, then every objection is an execution problem. This reframes every “we can’t do this” into “we haven’t figured out how to do this yet.”

Sidenote: I AM LITERALLY ONE DEGREE OF SEPARATION FROM ELON FKN MUSK WHAAAAT!

Yes yes, I understand proximity is not access. And access is not influence. The real question is: what do I do with this proximity?

3. The Personalized Medicine Play

I went to Mountain View for an event at the Google DeepMind office. After the event, I spoke with one of the DeepMind research scientists. I told him about my journey, what I’m building, why I’m in SF. He told me about a company called Diadia Health and connected me with the founders.

The core insight: If two people both catch a common cold, their bodies respond very differently. Their DNA is different. Their physiology is different. Their metabolism, hormone levels, genetic variants: all different. And yet, both get prescribed the same paracetamol at the same dose.

Diadia Health’s thesis is that medicine should be personalized: not at the level of “here are your symptoms,” but at the level of your actual genetic makeup. Their AI platform analyzes hundreds of biomarkers and generates personalized treatment protocols.

They focus on root causes, not symptoms. And the reason this works now is because AI can reason across millions of data points simultaneously: something no clinician, no matter how skilled, can do alone.

The founding story: The CEO is Elena Ikonomovska, PhD: former Google and Reddit AI scientist (she was Reddit’s first data scientist, later Head of AI at Change.org). She didn’t start Diadia because she saw a market opportunity on a spreadsheet. She started it because it happened to her. She had unexplained symptoms, her lab results kept coming back “normal,” and she spent years being dismissed by doctors. When she finally understood her own genetic profile, everything changed. Standard lab reference ranges were established decades ago using predominantly male subjects. They don’t account for genetic individuality, especially in women.

Her co-founder and CTO is Andrii Yasinetsky: ex-Uber, ex-Google, serial technical founder. They’d already co-founded a company together before (Mnemonic), so the team risk was de-risked before Diadia even started.

The distribution: They’re going B2B2C. They sell to functional medicine clinicians as a clinical decision-support tool: not directly to patients. Practitioners in functional medicine already believe in root-cause analysis, so initial adopters don’t need to be convinced of the philosophy, just the tool. They’ve validated across 12+ clinical sites and thousands of patient cases. Now they’ve also opened a direct-to-consumer path where women can order their own lab work and share reports with providers.

The Lessons

Founder-problem fit is real. Elena didn’t pick “personalized medicine” from a market study. She lived the problem. She was the patient who was dismissed. That kind of conviction doesn’t come from TAM analysis: it comes from rage and personal experience.

The co-founder question matters. Elena and Andrii had already built and shipped a company together before Diadia. They knew each other’s working style, conflict patterns, and technical strengths. In a space as complex as genomics + AI + healthcare, you can’t afford a co-founder learning curve.

Pick a wedge, not a market. Diadia isn’t trying to be “personalized medicine for everyone.” They started with a specific, underserved niche: women with complex hormonal conditions who keep getting told their labs are “normal.” That’s a wedge with real emotional urgency and a clear expansion path.

Distribution through believers. Going to functional medicine practitioners first: people who already think in terms of root causes: is a masterclass in picking your initial channel. You don’t waste energy convincing skeptics. You arm believers with better tools.

The Meta-Lesson

Three encounters. Three completely different domains: space, autonomous vehicles, healthcare. Same pattern:

Take asymmetric, contrarian bets. LongShot isn’t building a cheaper rocket. Tesla didn’t build a better LIDAR system. Diadia isn’t building a slightly improved blood test. Each of them looked at the dominant approach in their industry and said: that’s the wrong axis.

Founder-problem fit beats market analysis. Elena lived her problem. Mike’s military background gave him a unique lens on launch systems. Dhaval’s love for robotics since childhood made the Tesla bet feel obvious to him. In every case, the founder’s personal history made the contrarian bet feel like the only bet.

San Francisco rewards showing up. I walked up to a stranger at a WeWork and ended up touring a space cannon facility. I stayed after a DeepMind event and got connected to a precision medicine startup. A friend introduced me to a guy who demos self-driving software to Elon Musk. None of this was planned. All of it happened serendipitously.

The density of ambition here is unlike anywhere else. Every lunch, every event, every random conversation is a potential door. I hope I keep walking :)

Week one of many. Let’s see what next week brings.