Why AI Means Founders Should be Frugal

Hey y’all — here’s today at a glance:

🔗 Houck’s Picks

My favorite finds of the week.

Fundraising

• A successful process to raise your pre-seed (Link)

• A few notes about investing in startups from a full-time VC (Link)

• Things to know when raising a pre-seed round (Link)

• Elad Gil on how the best companies raise money (Link)

• Why founders shouldn’t let their rounds be “almosts” (Link)

Growth

• Controversial takeaways from not finding product-market-fit (Link)

• Step-by-step on how to go from $0 to $5M ARR (Link)

ICYMI

• A great name for your startup is more important than you think. Go to Kernel if your startup name is holding you back.*

• Why operating in grey area markets is a moat (Link)

• A definitive list of startup pivots (Link)

• How to attract and hire undiscovered talent (Link)

• What to know if you’re building the next billion-dollar marketplace (Link)

• Everything you need to know about startup valuation (Link)

• Why it’s so important to get an extremely precise persona of your user (Link)

💡 Opportunity: AI Custom Podcast

OpenAI continues to impress, and yesterday’s launch of GPT-4o (unfortunately the worst-named AI model yet) raised the bar once again.

A simple product that could do well would learn your interests and compile a daily podcast for you of a length you determine.

Going to workout for 2 hours? Here’s the news but with serious depth and backstories on the topics you’re most interested in.

Have a daily commute? Here’s your daily digest.

The new model would even allow for this to easily be a “conversation” where you tell to expand on or skip topics in real time.

Would likely make sense as either a simple subscription or usage-based model.

🧠 Framework: MVP

Found this straightforward way of thinking about what goes into an MVP on X last week.

The foundation of any MVP is a problem, but it’s also critical that the user is aware of this problem.

With an MVP you don’t want to have to spend any time educating the market — you want the lowest friction path to learning about the problem. You haven’t committed to solving the problem yet yourself.

The MVP becomes more attractive to build if the user has been actively looking for (or ideally hacked together themselves) a solution to solve the problem. This tells you it’s painful enough for them to likely pay for.

If they’ve already set aside the necessary budget to pay for it, that’s a winner.

📈 Trend: Retrieval Augmented Generation

Hallucinations seem to be a lingering risk for the usefulness of LLMs.

But reducing them just by increasing the scale of foundation models may be hard to continue to do.

As this very thorough analysis about how much text-based data literally exists and has not been scraped yet outlines, we may need to start optimizing through creative techniques rather than scale.

RAG is one of those techniques. It lets models reference external data sources before creating a response rather than relying only on its own training data.

💬 Quote: Frugal Founders

There’s a hidden problem in venture capital right now:

LLMs are enabling founders to do more with less, and founders are opting to take venture capital later or not at all.

Sam Altman says that AI agents will allow for the creation of a billion dollar, one-person company.

A big reason founders raise venture is to attract talent and support payroll earlier than revenue can, in order to hit milestones and grow.

If those growth milestones can be reached by a single person, they’re significantly more likely to be able to be supported by revenue alone, thus reducing the need for outside capital.

According to Paul Graham, this aligns with how founders should approach spending the money they raise anyway:

But… why would a founder raise venture if they’re able to build a company this way?

It would be defensive.

VCs want to deliver returns to their LPs. If they hear about or see a good opportunity, but that founder isn’t open to fundraising, they may just go to a founder they’ve worked with on a company before and fund them to build the same idea.

Giving the best VCs upside in a good idea may reduce your ownership stake, but it can also protect you from legitimate and well-funded threats.

💡 How I Can Help

🚀 Advertise in this newsletter to get in front of 80,000+ founders.