If you're a Fortune 500 company with a question about your business, you have options. Palantir will build you a $50 million integration platform. Accenture will send a team for two years. Snowflake will sell you a data lake and a six-month implementation.

If you're a 50-person company with the same question, your options are: ask the person who's been here longest and hope they remember, or export everything to Excel and start vlookup-ing.

This gap is absurd. Small businesses have data—often lots of it. They have questions that data could answer. They just don't have access to the tools that make it happen.

I think AI is about to change that. And I'm betting my company on it.

The Economics Changed

Building a "data platform" used to require:

This made economic sense for enterprises. A million dollars in infrastructure could save ten million in better decisions. But for a company doing $5M in revenue? The math never worked.

AI changes the economics in three ways:

1. Mapping happens in weeks, not months. The tedious work of understanding how "customer" in the CRM relates to "account" in billing—AI can accelerate this dramatically. What used to take a consulting team three months now takes a few focused weeks.

2. The interface is natural language. No query builder. No SQL training. Ask a question in English, get an answer. The "last mile" of data access—turning insight into something an executive can actually use—just got trivially easy.

3. Context is the only real cost. The AI models themselves are commodity. What makes them useful is context—understanding your specific business, your data, your terminology. Building that context is still real work, but it's a fraction of what "data transformation" projects used to cost.

Two Buyers, Same Product

As I've talked to potential customers, I've noticed two distinct buyer personas:

Efficiency Buyers want to save money. They're doing $20M+ in revenue, they know they're leaking value somewhere, and they want to find it. Classic ROI sale: "You're spending $200K/year on support that shouldn't exist. Here's where it's coming from."

Credibility Buyers want to look sophisticated. They're raising a Series A, or pitching enterprise clients, or preparing for acquisition. They need a slide in their deck that says "AI-powered business intelligence" and a demo that makes investors go "holy shit."

The beautiful thing: the same technology serves both. The ontology, the natural language interface, the connected data—it works whether you're optimizing operations or impressing stakeholders.

Credibility buyers are often faster to close (2-4 weeks vs. months) and smaller deals ($15-25K setup). Efficiency buyers take longer but pay more ($50-100K+). And here's the kicker: credibility buyers often become efficiency buyers once they realize the tool actually works.

Why Not Just Use ChatGPT?

Every small business owner has tried: "I exported my data to ChatGPT and asked it questions." Sometimes it works. Mostly it doesn't.

The problem is context. ChatGPT doesn't know that your "customer status" field has seven values, or that "churned" means something different for monthly vs. annual contracts, or that your sales team counts demos differently than your marketing team counts leads.

Without that context, you get plausible-sounding answers that are subtly wrong. Which is worse than no answer at all, because you might act on them.

The value isn't in the AI model—everyone has access to the same models. The value is in the context layer: the business ontology that makes AI actually useful for your specific business.

The BI-First Wedge

Originally, I thought the entry point would be code. I've spent 25 years maintaining business systems. I know how to make legacy software work. So naturally, I assumed: start with code maintenance, upsell to business intelligence.

I had it backwards.

Business intelligence is the easier entry. It's faster to demonstrate value (Week 1: "Here's what's in your data"). It requires less trust (read-only access, no code changes). And it creates the relationship that makes code work natural later.

"Your Stripe integration is dropping webhooks" is a much easier conversation when you're already in their systems showing them insights. Now fixing integrations seems obvious. Then maintaining code. Each step: more access, more trust, more revenue.

The code tools—the ontology, the governance layer—are infrastructure for the BI product. They're the expansion, not the entry point.

The Bet

Here's what I believe:

I'm building zeros to test this thesis. A "mini-Palantir" for businesses that can't afford Palantir. Business intelligence that actually answers questions instead of showing charts. AI that understands your business, not just language.

Maybe I'm wrong. Maybe small businesses don't actually want this—they're used to spreadsheets and gut instinct. Maybe the context problem is harder than I think. Maybe the economics still don't work.

But I've watched businesses struggle with the same data problems for 25 years. The technology to solve them finally exists. Someone's going to capture this market.

I intend for it to be me.

The Opportunity

Every small business with a database and unanswered questions is a potential customer. That's a lot of potential customers. The winners will be whoever figures out how to deliver enterprise-grade insight at SMB-grade prices. The technology is ready. The question is execution.