Pieter Levels Is Right About AI Slop. Here's Where the Real Pain—and Money—Is Hiding.

Forty-seven thousand dollars. That's the median annual revenue loss from subcontractor delays at a midsize US construction firm, according to a composite analysis of pain points our platform tracks. The root cause? Disorganized scheduling, last-minute change orders, and zero predictive intelligence about which subcontractors will show up late.

Now ask yourself: how many of the 80-plus seed-stage AI applications Pieter Levels reviewed recently could solve that? His viral post paints a picture of polished sameness—websites that are 8/10, copy that's grammatically flawless, MVPs that are "in principle fundable." But dig deeper and the themes blur together: AI chatbots, content generators, internal automation tooling. Levels calls it "slop," and for a large chunk of the market, he's not wrong.

Yet writing off all AI as bland misses a critical, data-backed story. While generic AI apps chase efficiency gains of 3.2 out of 5 severity, there's a whole other tier of problems sitting at 4.3, 4.5, even 4.8—and almost nobody is addressing them.

The Commoditization Problem Is Real

Levels' anecdote resonates because it aligns with broader patterns. Across 847 AI-related problems tracked on our platform, 61% fall under efficiency and automation use cases. That's a massive bucket, and it's where most founders are playing. But the average severity of these problems is just 3.2 out of 5. They matter, but they're not keeping operators up at night.

What's aggravating the sameness is the tooling environment. Foundation models and APIs make it trivial to spin up competent AI features, which is why Levels saw consistently decent MVPs. But competence isn't differentiation. As he noted, "the differentiating element was very hard to find." When anyone can build a decent chatbot or a document summarizer, the only moat is distribution or brand—which seed-stage startups rarely have.

This commoditization has a real cost for founders chasing the same playbook. Investors see through it. If you're building in the crowded center, your TAM slide might as well be wallpaper.

Where Builders Are Blind

Here's where the map gets interesting. Our data doesn't just confirm the AI slop trend—it reveals a sharp divergence.

Construction, an industry with 187 distinct problems tracked, carries an average severity score of 4.5 out of 5. Manufacturing (142 problems) clocks in at 4.3. Logistics (98 problems) at 4.1. These are not minor inefficiencies; they're operational nightmares with real-dollar arithmetic behind them—delay penalties, material waste, safety incidents. And yet, the number of AI app ideas targeting these sectors is shockingly low. Only a handful of solutions we track are purpose-built for these industries.

Part of the reason is domain access. These aren't industries a solo founder can grok in a weekend. They're thick with regulation, complexity, and tenured subject-matter experts who distrust outsiders. But that's exactly why they're defensible.

Consider the dynamic at play: while a wave of founders builds AI for other founders, the workers who pour concrete, manage warehouse inventory, or optimize CNC machines are using clipboards and WhatsApp. The pain is screaming, and the willingness to pay is high—if the solution respects the workflow.

The Revenue Blindspot

Levels also noted he hasn't seen anyone make something that actually increased revenue. That's a fair observation, but our data suggests the problem isn't that revenue-boosting AI doesn't exist—it's that it's underrepresented in the startup pipeline. Out of the 847 AI problems we track, only 8% of app ideas target revenue generation directly. But those that do carry a severity rating of 4.1 out of 5, indicating operators feel the absence acutely.

Dynamic pricing engines, sales forecasting, supply chain optimization—these are not slop. They require domain-specific data, modeling rigor, and integration with legacy systems. The problem space is there. What's missing is founder attention.

Where the High-Stakes Problems Live

If you're an investor or a builder skeptical of AI hype, the signal is in the severity gradient. On our platform, the highest-severity AI problems cluster around industries where accuracy and trust are non-negotiable. Legal document analysis: severity 4.6, with clear feature gaps around domain specificity and explainability. Healthcare diagnostics: severity 4.8, with only 12 app ideas addressing it. These are not markets you win with a thin wrapper around a general-purpose LLM. They demand precision, compliance, and deep workflow integration—and that's precisely why they'll have pricing power.

Meanwhile, the internal tools Levels described—server monitors, backup checkers, auto-bug-fixers—are useful, but they're also the ones rapidly becoming commodity utilities. Their severity scores trend lower (2.8–3.3 range in our data), and their value is in time savings, not in generating new revenue or preventing catastrophic loss.

The Next Wave of AI Value Creation

Levels ends his post wondering if the groundbreaking stuff comes next year. The answer depends on where founders and funders aim their attention.

If the market stays mesmerized by the efficiency bucket, we'll get incremental slop. But if builders look at the severity map—at construction, manufacturing, logistics, legal, healthcare—they'll find problems that aren't just fundable, but defensible and deeply valuable.

For investors, the takeaway is clear: the next AI unicorns won't look like polished MVPs for content generation. They'll look like unglamorous tools solving unsexy, high-stakes problems in industries that haven't had a tech refresh in decades. The slop wave was a warm-up. The real build starts now.

At PainSignal, we track where market pain lives so founders and investors can see what's worth solving. The data is clear: the biggest opportunities in AI are hidden in plain sight, behind industry gatekeepers and the "boring" problems nobody wants to tweet about. While the world builds slop, those who dare to get vertical will define the next decade.