Not Everyone Can Build an AI VP of Finance (And That's the Opportunity)

You can read Jason Lemkin's latest piece on SaaStr about building an AI VP of Finance and feel one of two ways. Inspired, because he connected bill.com, QuickBooks, Brex, and PandaDoc in under two hours and cut contract-to-invoice from a day to 30 seconds. Or slightly panicked, because your own collections process still involves a spreadsheet, three reminders, and a prayer.

The gap between those two reactions is exactly where this story gets interesting.

Lemkin's thesis is that agents are collapsing into each other. Instead of 100 narrow AI tools, his stack is converging into fewer, deeper agents that share a common knowledge base. His AI VP of Finance isn't a standalone app—it runs inside 10K, the same agent handling marketing. And it works. It reads contracts, creates invoices, updates Salesforce, and even started calculating commissions on its own. The man runs SaaStr with three humans and twenty-plus agents in production.

But here's what his article doesn't address: most businesses aren't even running one agent yet.

According to our data, which tracks thousands of small business pain points, 72% of app ideas focus on single-function automation tools. Think "automated invoice reminders" or "AI customer support agent." Not multi-agent orchestrators. Not collapsed monoliths. Simple, narrow solutions that solve one problem well. The average severity of accounts receivable problems across 15 industries is 4.2 out of 5. That's a lot of pain. And the most common complaint isn't "I need a VP of Finance agent"—it's "my invoicing software doesn't talk to my CRM."

We track 22 distinct problems about data silos and integration failures, with a severity of 4.0. These are companies stuck on the fundamental step: getting one tool to pass a customer name to another. They're not ready for a monorepo of agents. They need a plug and play pipeline.

The author's approach makes for a great north star. It shows what's possible when you have technical chops, a lean team, and a willingness to vibe code your own integrations. But for the other 90% of businesses—the dentist office with overdue invoices, the construction firm chasing payments on 7 different fronts, the professional services agency using three separate systems—the priority is different.

Our data also shows that many companies start by automating low pain workflows first, precisely because those are easier to implement. They build confidence before tackling the high pain areas like collections. That's a valid strategy too, even if it's the opposite of Lemkin's "start where it hurts" mantra.

So what's the real opportunity?

It's building those pre-integrated, single-function agents that work out of the box. The AI that sends payment reminders without needing to touch an API. The tool that reconciles invoices from QuickBooks and Brex without a developer account. The agent that reads a contract and creates an invoice, period—not one that also calculates commissions and generates new contracts, but one that does its one job flawlessly.

Lemkin's experience is proof that the future is agentic. But the present demand—the one hiding in our data—is for accessible, low friction automation that works for a plumber in Texas as easily as it works for a SaaS CEO in Silicon Valley. That's where the next wave of successful tools will come from.

So by all means, read the piece and dream big. But while you're dreaming, remember that 47 problems in our database mention API integration difficulties, averaging 3.8 out of 5 in severity. The gap between SaaStr and the rest of the market isn't a technology gap. It's a simplicity gap. And that gap is an opportunity dressed in plain clothes.