AI Pilots Are Failing

Hey y’all — here’s today at a glance:

🔗 Houck’s Picks

My favorite finds of the week.

Fundraising

• Dan Swygart on what the process of raising VC money actually looks like (Link)

• Ilya Strebulaev on which angels have backed the most unicorns (Link)

• Andrew Yeung on who every founder should have on their cap table (Link)

• Ben Cashnocha on what late stage firms are telling seed stage founders (Link)

Growth

• Jesse Pujji on how to rebuild a $20M business to 8-figures in 18 months (Link)

• Natia Kurdadze on lazy content marketing for startup founders (Link)

• Denislav Jeliazkov on how users only judge your entire app by 2 moments (Link)

ICYMI

• This is how decision-makers read less but know more (Link)

• Anshuman Sinha on co-founder equity split strategy (Link)

• Blake Scholl on everything we’ve been taught about burnout is wrong (Link)

• Z Fellows on Naval Ravikant’s advice on how to build a team that ships (Link)

💡 Opportunity: AI-Safe Content Shield

I don’t have any kids yet but, if I did, I’d do everything to keep them off the internet.

Sounds crazy, maybe (after all, the internet enables me to make a living and connect with friends all over the world while staying up to date on bleeding edge tech and where the world is headed) but it’s the truth.

There’s simply too much risk, even for me.

Over 54% of kids have been exposed to porn by 13 years old.

Social media use among adolescents has been tied to depression.

And AI, through deepfakes and more, has made protecting your kids online more challenging than ever.

Regardless of where you stand on this as an individual, it seems reasonable to assume there’s a large and growing market of parents who would be looking and willing to pay for a new, AI-powered generation of tooling to push back against this.

A “content shield” that’s assessing what’s about to be shown on the screen of your kid’s phone or computer before it is, and filters out anything that doesn’t meet your guidelines seems like an easy sell.

These types of tools existed pre-AI but aren’t equipped to handle AI-generated content.

🧠 Framework: The CLEAR Pitch Framework

First-time founders persuade.

Second-time founders educate.

With respect to Theranos, FTX, and others, investors aren’t going to be "fooled” into investing in your startup.

You have to clearly communicate a venture-scale opportunity to them.

When finalizing your pitch deck, and analyzing any investor convos after the fact, grade yourself through these five lenses:

• C → Did you provide CLARITY on the market and problem?

• L → Were your numbers and vision based in LOGIC?

• E → Did you offer EVIDENCE that people care about your solution?

• A → Did you come off as AWARE of the risks?

• R → Are you READY to act on an opportunity right after the investment is made?

Remember — you’re not selling a dream, you’re making a decision possible.

📈 Trend: AI Pilots Are Failing

If you’ve been holding out on implementing anything more than paying for ChatGPT subscriptions for your employees, maybe keep holding out a bit longer.

At least that’s what a new report out of MIT says.

Among the companies surveyed, while 80% of “general-purpose LLM" pilots (aka ChatGPT, etc) have led to a rollout, only 25% of pilots for more task-specific AI tools have.

Furthermore, the report claims that even thought the general-purpose rollouts are improving “individual productivity” that isn’t consistently translating to P&L improvements.

The report characterizes this more pessimistically by saying that this means AI is failing “95%” of companies. That’s a bit of an odd takeaway, since even the more challenging “task-specific” AI tools are only being piloted at 20% of companies to begin with.

Either way, the root cause appears to be that “most GenAI systems do not retain feedback, adapt to context, or improve over time.”

Basically, these tools are still like junior employees — except they have amnesia.

The opportunity: it would be challenging if not impossible to outcompete OpenAI, Anthropic, and xAI on model quality at this point, but there may be breakthroughs in memory that you can build infrastructure tooling around. Who knows, maybe you’ll get acquihired.

And, if you are building an AI-powered tool, emphasize persistence and memory in your system prompts ad nauseam.

💬 Quote: Communicate Your Vision

Now more than ever, the right ideas has real value.

Execution is no longer everything — AI is closing the gap slowly but surely from month to month.

Maybe it never was.

Sam Altman even says that picking the right thing to work on in the first place is the most crucial element of productivity.

Regardless, just having the right vision won’t mean much if you don’t do anything with it.

(And, with tools like Idea Browser, good ideas are cheaper than ever too.)

But even if we completely discount executing on the idea, there’s a step to master that’s more important than the idea itself — how you communicate it.

Think about it.

Founders are selling their idea everywhere they go. To potential investors, co-founders, employees, partners, users, friends, family, etc.

The best leaders prioritize clear communication because they know that without it, no one will follow them.

And founders, in particular, have an exceptionally high bar to climb over when selling their idea because the idea is inherently brand new and disruptive.

Founders need to convince people not only to want to support them, but that what they are doing differently than expected is right and will succeed.

If you’re a founder who’s had trouble rallying one or more of those groups I mentioned on getting behind you, audit how you’re talking about your vision to them specifically.

It might make all the difference.

How We Can Help

Advertise in this newsletter to get in front of 75,000+ founders.