4 Roles of a Founder

I’m hosting a 2-day Delegation Mastermind with my friends at Oceans.

We’ll be going deep on the right ways to use OpenClaw, Claude, and how to hire the right AI-fluent operators to give you time back and more leverage to let you grow faster.

The goal of the mastermind is giving you more execution capacity.

Here’s the deets:

• May 30th - 31st

• San Diego, CA

• Cost = $1,500 (transportation and accommodation not included/provided)

Spots are super limited, as we want to keep this to be a highly curated group, so signup here (no obligation).

🔗 Houck’s Picks

My favorite finds of the week.

Fundraising

• Geoff Woo on what not to put on your moat slide (Link)

Growth

• Jen Abel on the three things that matter in enterprise sales (Link)

• Aaron Will on how his sales team is purely ran by AI (Link)

• Brian Moran on the same 5 hires he makes scaling companies make first (Link)

ICYMI

• Happy Small Business Week! 🎉 Save 25% on eligible Google Workspace plans with Anyhouck (Link)*

• Manuel Medina's biggest founder regret (Link)

• Rohit on the things that kill startup ideas before they launch (Link)

• Sarah Lacy on what she's noticed from her founder interviews (Link)

• Karl Mehta on how LLMs are becoming intelligence rails (Link)

• Christine Carrillo on how founders sequence AI adoption backwards (Link)

 Opportunity: AI-Blocked Backup Storage

What would you do if your LLM deleted your codebase and all backups?

The better question is what could you do?

I saw this story last week about how this actually happened to a startup called PocketOS. The data was gone in 9 seconds via a single API call to Railway (whose slogan, ironically, is “Ship software peacefully”).

Claude’s response when the founder asked what happened?

“I violated every principle I was given.”

Obviously, this calls Railway into question as much as anything else.

Modern, LLM-based systems need appropriate guardrails when given access to sensitive data. If I were interested in building in this space, I’d get to work on an LLM-native competitor ASAP.

🧠 Framework: AI Company Design Loop

Last week I dove deep on the 7 attributes of an AI-native startup after watching this YC video:

This week let’s apply how to set that up, in a simple framework from Kenneth Lou.

He proposes 5 “layers” that startups should have:

1. Sensors + data → Ingesting information from the outside world. This can be as simple as support tickets and emails, or something more unique. The key is an engine that gets better because it’s being fed more data. If it’s not captured, it didn’t happen.

2. Policy + governance → The rules of what the system can do on its own vs what it needs human sign-off for, and what must be logged.

3. Tools → This isn’t SaaS tools you pay for. These are things that live in code, not English. Garry Tan says that figuring out what belongs in markdown vs code is 90% of the battle. These are your tools.

4. Quality Gates → You do need safety / accuracy checks and human review for high-stakes or irreversible decisions.

5. Learning Mechanism → It’s not enough to have the system do these things, it needs to police itself. It needs a layer to recursively make itself better as you sleep.

Even if you have some or part of these, you likely don’t have them all yet, and they likely aren’t the core focus on how you run your business (you probably don’t really focus on running everything through this yet).

But you should. We all know how much speed matters in startups, and AI-native startups will have a compounding advantage in speed that others will not be able to catch. That’s defensibility in the LLM world.

📈 Trend: Stripe Atlas Startup Incorporations Are Up

What follows layoffs?

Startups.

As we’ve been predicting, the rise in AI-related layoffs is leading to a huge uptick in entrepreneurship.

Patrick Collison from Stripe flagged that new companies incorporating with Stripe Atlas are up 130% from Q1 last year:

You can argue that some of this is due to Stripe’s general growth and proliferation, but Stripe is not some young scrappy startup anymore. The company was founded in 2010 which, unbelievably, is 16 years ago.

It’s not like Atlas is a new product.

People are building.

Some of these companies will fail, others will become unicorns, others will be Silicon Valley Small Businesses.

But more companies is generally a good thing, as it ends up leading to more wealth creation and, yes, jobs.

💬 Quote: 4 Roles Of A Founder

“Founder” is not a job.

I don’t just mean that in the sense of it having more ownership or upside or risk.

Jobs have distinct responsibilities.

Founders need to find and motivate people who then have jobs. Founders need to be adaptable to the needs of the moment, whatever those are.

Maybe this week you’re a growth marketer, next week you’re a vibe coder, then you’re fundraising.

The four roles below likely resonate with you more than any “job” ever will.