AI's Real Rewiring: It's Not Just Sales, It's Every Broken Workflow

A preschool art teacher is trying to manage a room of four-year-olds. One kid is having a meltrum over a broken crayon, another is about to spill glitter everywhere, and the teacher’s clipboard—where she’s supposed to log behavior incidents—is sitting empty on a shelf. She’s not thinking about AI. She’s thinking about survival.

This isn’t a sales floor. It’s a classroom. But the underlying problem is the same one that Greg Beltzer at Salesforce described when he talked about messy data and processes that live in people’s heads. The teacher’s knowledge—which kid is prone to tantrums, which activity triggers chaos—exists in her memory, not in a system. It’s unstructured, it’s lost when she’s overwhelmed, and it never makes it into a record that could help her or her successor.

Our data at PainSignal tracks 2,784 operational problems across 92 industries. And what jumps out isn’t the dominance of sales challenges—it’s the sheer volume and severity of pain in places like education, healthcare, and retail. In education alone, we have 47 problems tracked, with an average severity score of 4 out of 5. Science teachers can’t run labs because equipment is broken and nobody logs the issue. Tutoring businesses lose clients because follow-up falls through the cracks. These aren’t niche complaints; they’re systemic failures that mirror the ‘unsexy problems’ Jason Lemkin writes about in his SaaStr piece, but with higher stakes and less tooling.

Lemkin’s article pulls together insights from Salesforce, Momentum, and Mangomint on how AI is rewiring B2B sales. The lessons are solid: start where your data is cleanest, use AI to follow up on neglected leads, fix your CRM mess, and don’t raise quota until you’ve built the systems. But reading it with our dataset feels like watching someone expertly diagnose a fever while ignoring the plague next door. The principles are universal, but the opportunities are unevenly distributed.

Take lesson two: ‘Every Company Has Leads They Never Followed Up With. AI Fixes This Immediately.’ Greg Beltzer admits that Salesforce had a ‘shockingly low percentage’ of inbound leads being followed up on. Reps cherry-picked, the rest got fake loss reasons. An AI agent on those orphaned leads generated revenue they wouldn’t have had otherwise.

Now, look at healthcare. We track 31 problems in this vertical. Pharmacists are drowning in prescription intake, with severity scores hitting 4/5. Missed follow-ups here aren’t lost revenue; they’re medication errors, patient dissatisfaction, and regulatory risk. The ‘leads’ are patients needing refills or consultations that slip through the cracks because the pharmacist is juggling five things at once. An AI agent that consistently logs inquiries, schedules call-backs, and surfaces urgent cases could save lives, not just quota. But how many AI startups are building for independent pharmacies instead of sales teams?

Or consider lesson three: ‘Your CRM Data Is Worse Than You Think, and AI Will Show You.’ Marchelle Mooney from Mangomint found her Salesforce instance was a mess—deals closed with no notes, marketing automation had more data than CRM. Her top rep closed 35 logos a month, but everything lived in text messages. Sound familiar? In education, we see the same pattern. Teachers’ gradebooks, behavior logs, and parent communications are scattered across Google Sheets, email threads, and paper notes. There’s no ‘CRM’ to even be messy. The first step isn’t deploying an AI agent; it’s deploying tools that capture data into a single source of truth. But in many of these industries, that foundational tooling doesn’t exist yet. The gap isn’t just dirty data; it’s no data.

This is where our data challenges the article’s framing, respectfully. Lemkin highlights retention as an ‘underrated play,’ with Ashley Wilson from Momentum pointing out that customer success teams are underserved by tools. True. But our data shows retention isn’t underrated as a problem—it’s a persistent pain point across industries. In education, tutoring businesses struggle with client retention and refunds (we track 11 problems in customer management). The gap is in effective AI tools tailored to specific retention scenarios. For example, an AI that listens for dissatisfaction signals in parent-teacher conferences or flags students at risk of dropping out could have more impact than a sales chatbot. Yet, as the article notes, ‘almost nobody is deploying there yet.’

So, what does this mean for you, the builder? If you’re a vibe_coder or indie_hacker looking for a project, the lesson isn’t to build another AI sales assistant. It’s to take these validated principles and apply them to a vertical where the pain is acute and the solutions are scarce. Start with our Workflow Automation category, where we track 20 problems—many in industries like education and healthcare. These are structured processes begging for automation, just like the service and ops workflows Greg Beltzer found easier for AI than sales.

Or, explore Education problems directly. You’ll see issues like ‘Science teacher unable to conduct labs due to broken equipment’ (severity 4/5) or ‘Preschool art teacher facing safety issues during tantrums’ (severity 4/5). These aren’t just complaints; they’re blueprints. An AI that monitors equipment sensors and auto-generates maintenance tickets could save lab time. A simple tool that logs behavior incidents via voice-to-text could give teachers data to prevent meltdowns. The opportunity isn’t in replacing the teacher—it’s in fixing the basics she never has time for, exactly as the sales leaders described.

For agency_devs who already serve these verticals, the insight is even more direct. Your clients in retail or fitness aren’t worrying about CRM data; they’re drowning in inventory mismatches, class scheduling chaos, and member churn. Our data shows retail problems like ‘clothing store owner struggling with size allocation’ or fitness issues like ‘gym manager overwhelmed by member check-ins.’ These are workflow inefficiencies with clear data inputs (sales logs, attendance scans) that AI could structure and act on. The playbook is the same: clean the data first, automate the neglected tasks, use AI to lift middle performers—but the context is a boutique, not a sales floor.

And for seed_investors, the pattern-recognition here is critical. The article’s lessons are market signals: AI wins when deployed on unsexy problems with clean-ish data. Our data maps where those conditions exist outside of tech. Industries like education and healthcare have high severity scores (4/5 average) and low tool saturation. They’re not yet AI-native, but their problems are ripe for the same rewiring. Investing in startups that apply sales-grade AI to these verticals isn’t a niche bet; it’s leveraging proven frameworks in underserved markets.

Ultimately, Lemkin’s piece is right: ‘The companies that will pull ahead aren’t the ones with the fanciest agents. They’re the ones that got their data right, deployed AI on the unsexy problems first, and invested in the operational infrastructure.’ But our data screams that those companies aren’t just in B2B sales. They’re in every industry where operational pain is severe and visibility is low. The rewiring is happening; it’s just starting where the sparks fly—in classrooms, clinics, and small businesses that have been ignored by the first wave of AI tools.

So, before you build another sales bot, take a look at the problems we track. You might find that the real AI revolution isn’t in closing deals, but in fixing the broken workflows that keep people from doing their best work. And if you’re looking for where to start, browse our problem marketplace—the data is waiting.