AI's Support Crisis Isn't Just Annoying — It's a $50M Opportunity

I stumbled on this piece from Jason Lemkin over at SaaStr about the state of customer support in AI apps. It's a raw, from-the-trenches take that'll resonate with anyone who's tried to get help from a trendy AI tool recently.

Lemkin's thesis is simple: most AI apps have either no support or support that's so broken it makes things worse. He walks through his own experiences with Get Recall (a broken upload after a release with zero recourse) and Clerk (a months-long odyssey to cancel a second account). It's the kind of story that makes you nod your head in shared pain.

But here's the thing. Lemkin frames this as a problem rooted in founder mindset — they don't think support matters, they think AI will solve it, or they're growing too fast to care. That's all true. But after spending time with PainSignal's data on this space, I think there's another layer: this is also a market inefficiency that's ripe for disruption.

PainSignal tracks user-reported problems across software tools. In the 'Communication' category alone — which covers support responsiveness, documentation clarity, and feedback channels — we've logged 40+ distinct problems for AI tools, with an average severity of 4.1 out of 5. That's not a few disgruntled users. That's a systemic pain point.

Compare that to the 'Onboarding' category, where we've tracked 25+ problems with an average severity of 3.5. Or 'Pricing' at 3.2. The data suggests that support-related issues are not just common — they're the most painful and least resolved. And this across a broad swath of AI apps, not just the ones Lemkin name-checks.

So why is this a market opportunity? Because the incumbents are ignoring it. If you're an indie hacker or seed-stage founder looking for a wedge in AI, building a company around solving the 'support gap' could be your entry point.

Think about it: there are thousands of AI apps out there, and the vast majority are run by small teams. They don't have the resources to staff a support team. They don't have the expertise to build a knowledge base. They don't have the time to train an AI bot. They need someone else to do it for them. That someone could be you.

Lemkin's advice to founders is solid — put a real email address on your site, hire a support person before your next engineer, read every ticket for 30 days. But that advice assumes you can afford to slow down. Most AI startups can't. They're racing to ship features and acquire users before the next funding round.

What if there was a service that handled support for a portfolio of AI apps? A kind of 'support as a service' that combines AI triage with real humans who understand the product category. You'd aggregate demand across multiple apps, amortize the cost of the human team, and offer each app a branded support experience. The AI apps get to focus on product. The users get actual help. And you build a business with recurring revenue that's largely uncorrelated with any single app's success.

I'm not saying this is easy. The unit economics are tricky at the $10–20/month price point that many consumer AI apps charge. But the PainSignal data shows that the users who are most vocal about support gaps are also the power users — the ones who'd likely pay more for a premium tier that includes real support. In fact, several of the documented problems mention 'Would pay extra for a support SLA.' That's a signal worth following.

Lemkin ends his piece by predicting that the AI apps that combine brilliant agents with brilliant support experiences will win. I agree. But I'd add that the infrastructure to deliver that support doesn't have to be built in-house. It can be a product in its own right.

So if you're an indie hacker looking for a niche, or an investor trying to spot patterns, look at the companies that are enabling the AI ecosystem to focus on their core product by handling the messy human layer. The data says the demand is real, the pain is deep, and the incumbents have left the door wide open.