Most conversations about AI in business start with the technology. What can it do? How fast can it run? What's the latest model?
But here's what I've learned building workflows: the AI is almost never the constraint. The data is.
I built a company research brief, the kind you take into a meeting. The AI was strong at synthesizing and spotting patterns. But I started with public data: the company site, recent news, filings. The stuff anyone can reach. The output was competent. Useful, but generic.
Then we fed it the premium sources, Bloomberg, PitchBook, Crunchbase, the ones that sit behind paywalls. Same model, same prompt. The brief came back specific. The kind of read a competitor on public data alone would miss.
That's the real constraint. Not whether the AI can think. Whether you have access to data that's worth thinking about.
I see teams get frustrated with AI because they're feeding it thin data and expecting thick insights. Public information. Surface-level internal records. Scattered context. And then they blame the AI for giving them obvious answers.
But the AI is only as good as what you give it.
The teams that get outsized value from AI are usually the ones who either have rich internal data, years of transaction history, detailed customer context, proprietary workflows, or they've invested in premium data sources that give them information their competitors don't have access to.
That's when AI becomes a real lever. Not because the model got smarter. Because you gave it something real to work with.
So before you evaluate an AI tool, ask yourself: what data do I actually have access to? Is it public information anyone can get? Or is it proprietary, detailed, rich with context? Can I subscribe to better sources? Do I have internal data I'm not using?
Because the constraint isn't the AI. It's the data feeding it.
Everything else follows.