A year ago I made an investment decision based on an AI company breakdown I saw on Twitter. The thread had 50K likes, the analysis looked solid, and the company sounded like a no-brainer.

I lost money on that one.

Not because the person was lying — they just didn't know what they didn't know. They were looking at surface-level metrics (revenue growth, TAM size, partnership announcements) without digging into the stuff that actually matters for AI companies specifically.

That experience sent me down a rabbit hole. I started reading every 10-K filing, every technical paper, every earnings transcript I could find from AI companies. And I realized something:

Evaluating AI companies requires a completely different lens than traditional tech.

Here's why:

• Technology risk is higher. A new open-source model can commoditize your entire product overnight. You need to understand where a company sits in the stack.

• Revenue quality varies wildly. "AI revenue" can mean anything from high-margin SaaS to low-margin compute reselling. The income statement alone won't tell you which.

• The hype premium is real. AI companies trade at multiples that only make sense if you believe their growth projections — and most of those projections are fantasy.

• Technical due diligence actually matters. You don't need to be an ML engineer, but you need to understand enough to ask the right questions. "How is this different from what OpenAI offers for free?" is a great starting point.

I started writing up my research and analysis because I figured other investors were dealing with the same problem. That turned into Our Cerebral Successors — a weekly newsletter that applies real technical and financial analysis to AI companies.

No hype, no pump pieces, no "this will 10x" nonsense. Just honest breakdown of what's real and what's noise.

If that sounds useful, it's 20% off right now with code EARLYBIRD. Or just follow along here — I share free insights on this forum regularly.

What's your experience been with AI investing? Any wins or losses you've learned from? Would love to hear below 👇

I've analyzed dozens of AI companies at this point, and certain patterns show up over and over when a company is more sizzle than steak.

Save this list. It'll serve you well.

🚩 1. They can't explain what their AI actually does

If the website, pitch deck, and earnings call all use vague language like "leveraging AI to transform outcomes" without ever explaining the actual mechanism — be suspicious. Real AI companies can tell you exactly what problem their model solves and how.

🚩 2. Revenue growth outpaces everything else

Revenue growing 200% while headcount grows 300% and margins are shrinking? That's not scaling, that's buying growth. Healthy AI companies become more efficient as they grow, not less.

🚩 3. All their customers are "pilots"

Pilots are great. But if 18 months in, the company still has 30 pilots and 2 production customers, something's wrong. The pilot-to-production conversion rate is one of the most telling metrics in enterprise AI.

🚩 4. The CEO has more Twitter followers than the product has users

Harsh but real. If the company's growth strategy is primarily "founder personal brand," that's a media company, not an AI company. Check the product metrics, not the podcast appearances.

🚩 5. They pivoted to AI in the last 12 months

A company that was doing something completely different 18 months ago and suddenly "pivoted to AI" after ChatGPT went viral? That's trend-chasing, not technology building. Real AI companies have years of R&D behind them.

None of these are automatic disqualifiers — but stack two or three of them together and you've got a high-hype, low-substance situation.

These are the kinds of patterns we dig into every week in the newsletter. Real names, real numbers, real analysis.

What red flags would you add to this list? 👇

Hey, thanks for stopping by 👋

If you found your way here, you're probably like me — excited about AI but exhausted by the hype.

Every day there's a new "revolutionary" AI company, a new $10B valuation that makes no sense, and a new prediction about how AI will change everything by next Tuesday. It's a lot.

Our Cerebral Successors exists to cut through all of that.

Here's what we actually do:

📊 Weekly deep dives — Real analysis of AI companies, their technology, their financials, and whether the hype matches reality. No jargon for the sake of it. Written for people who make actual investment decisions.

🔍 Earnings breakdowns — When AI companies report, we pull apart the numbers and tell you what they mean. The stuff that doesn't make it into the headlines.

🚩 BS detection — We call out misleading AI claims, inflated revenue projections, and companies that are "AI-powered" in name only.

💡 Under-the-radar finds — The companies nobody's talking about yet that probably deserve your attention.

This public forum is where we share free insights, talk shop, and have real conversations about AI investing. The full weekly newsletter with deep dives and analysis is available with a subscription.

If you're an investor or fund manager tracking AI — you're exactly who this is for. Use code EARLYBIRD for 20% off your first 3 months.

Feel free to ask questions, share companies you're tracking, or just say hey. This is a community, not a lecture hall.

Let's cut through the noise together 🧠

I talk to a lot of investors who tell me some version of: "I know AI is important, I know I should have exposure, but I have no idea how to separate the real ones from the noise."

Fair. It's genuinely hard. So here's the framework I use — feel free to steal it.

The 5-Question AI Company Filter:

1. What's the technical moat?

Not "we use AI" — that's not a moat. I mean: do they have proprietary training data? A unique architecture? A feedback loop that makes their model better the more it's used? If the answer is "we fine-tune open source models," that's worth investigating further. Barriers to entry matter more in AI than almost any other sector because the underlying technology is commoditizing fast.

2. Where's the margin?

Compute is expensive. If a company is growing revenue but margins aren't improving, they might just be subsidizing usage to hit growth targets. Look at gross margin trends over 3-4 quarters minimum. Healthy AI businesses show margin expansion as they scale.

3. Who's actually paying?

Enterprise contracts > consumer freemium. Look for evidence of real enterprise adoption — named customers, case studies with specific ROI numbers, multi-year contracts. "1 million free users" means nothing if none of them pay.

4. What happens when GPT-5 drops?

Seriously ask this. If a new foundation model release could make this company's product irrelevant overnight, that's a problem. The best AI companies are building on top of foundation models, not competing with them.

5. Is management technical?

In AI, this matters more than other sectors. You want a CEO or CTO who can actually explain their technology without buzzwords. Read their interviews, listen to podcast appearances. If every answer is vague hand-waving about "leveraging AI to transform industries"... dig deeper.

This is a simplified version of what I cover in much more depth in my newsletter Our Cerebral Successors. Every week I apply this kind of analysis to specific companies, sectors, and trends.

20% off the first 3 months with code EARLYBIRD right now.

What would you add to this framework? Curious what other filters people use 👇

I feel like 90% of AI discourse right now is about five companies. You know the ones.

Meanwhile there's this entire layer of AI businesses that most people have never heard of — and they're genuinely profitable. Not "we'll figure out monetization later" profitable. Actually profitable. Right now.

I've been spending most of my time researching these kinds of companies, and a few themes keep coming up:

Infrastructure picks-and-shovels plays are crushing it

Everyone talks about the model makers. Almost nobody talks about the companies building the boring stuff — data labeling, model monitoring, AI security, inference optimization. These businesses have real recurring revenue, actual margins, and way less existential risk than trying to build the next foundation model.

Vertical AI is where the real money is

The companies applying AI to a specific industry problem (not trying to be everything to everyone) are the ones with the stickiest customers. Think AI for radiology reads, AI for insurance claims processing, AI for materials science. Narrow focus = deep moats.

The "boring AI" advantage

If a company's AI product isn't flashy enough to go viral on Twitter, that's actually a good sign. The less sexy the application, the less likely a big tech company is going to come destroy your market. Nobody at Google is staying up at night thinking about AI-powered freight logistics optimization.

I write about companies like these every week in my newsletter — the ones flying under the radar that actually deserve attention from serious investors. It's called Our Cerebral Successors and it's basically the anti-hype AI investment newsletter.

If this resonates, come check it out. There's a 20% discount right now with code EARLYBIRD.

But more importantly — who else are you watching in the "quiet AI" space? I'd love to hear what's on your radar. Drop it below 👇

Hey everyone 👋

So I've been doing something kind of obsessive for the past several months. I've been pulling apart earnings calls, investor presentations, and quarterly filings from basically every publicly traded company making serious AI claims.

Here's the thing that keeps jumping out at me:

Most "AI revenue" isn't what you think it is.

A huge chunk of what gets reported as AI revenue is actually just repackaged cloud computing spend. Company announces an "AI product," investor slides show hockey-stick projections, stock jumps 8%... and then you dig into the 10-K and realize their "AI solution" is a thin wrapper around an API they don't even own.

Three patterns I keep seeing:

1. The "AI-Powered" relabel

Take an existing product, add a chatbot or recommendation engine, call it AI-powered. Revenue stays the same but now it's "AI revenue." I've seen at least a dozen companies pull this in the last two quarters.

2. The pilot-to-production gap

Company announces 15 enterprise pilots. Sounds amazing. But pilots convert to production contracts at maybe 20-30%. And the production contracts are often 60% smaller than the pilot values that got announced. The press release math never adds up.

3. The gross margin tell

This is the one most people miss. True AI-native companies should be improving margins as they scale (models get more efficient, inference costs drop). If margins are flat or declining while "AI revenue" grows, they're probably just reselling compute at thin margins.

I started writing about this stuff because I genuinely couldn't find anyone breaking it down in plain english for investors. No PhD required, no jargon for the sake of jargon — just real analysis on what's actually happening.

That's basically what I do with my newsletter Our Cerebral Successors — weekly deep dives on AI companies, cutting through the marketing to find what's real. If you're making investment decisions around AI, it might save you from a few expensive mistakes.

Currently running 20% off with code EARLYBIRD if you want to check it out. But honestly, even if you never subscribe — save this post. These three patterns alone will change how you read the next AI earnings call.

Happy to discuss any of this below 👇

Most analysts covering AI companies are using the wrong metrics. Here's what actually matters:

1. Inference cost per query, not training cost

Everyone obsesses over training runs. But the companies that win will be the ones who drive inference costs to zero. Watch cost-per-query trends, not capex headlines.

2. Retention cohorts, not MAU

AI products have the worst retention in tech history. A tool with 100M signups and 5% 30-day retention is worth less than one with 500K users and 60% retention. Always ask: are people coming back?

3. Revenue per API call

The picks-and-shovels play in AI isn't chips anymore — it's inference APIs. Track revenue per API call across providers. The spread tells you who has pricing power.

I write about this stuff every week in Our Cerebral Successors. Deep dives on the AI companies and technologies that actually matter for your portfolio — not the hype cycle.

If you allocate capital anywhere near AI, this is your edge.