Two professional workflows can be nearly identical in structure — high document volume, high stakes, recurring deadline pressure, clear output requirements — and have completely different competitive dynamics. One might be wide open. The other might already have a well-funded incumbent with multi-year contracts at the firms you’d be selling to. The AI capability required is the same. The market situation is not.

This asymmetry shows up when you look at adjacent professional verticals that seem similar from the outside. Real estate due diligence and legal due diligence for M&A transactions, for instance, share obvious surface similarities. Both involve reviewing large volumes of complex documents under time pressure before a high-stakes transaction closes. Both require domain knowledge to extract what matters. Both have workflows that currently span multiple disconnected tools.

But the competitive landscapes are radically different. The legal M&A space has been targeted by well-capitalized AI companies for years. Large law firms — the buyers in that vertical — already have enterprise AI relationships in place, are risk-averse about switching, and have compliance and confidentiality requirements that make evaluation cycles long and switching costs high. Entering that market means competing against incumbents with deep deployment histories at the exact firms you need to win.

Real estate acquisition teams don’t look like that at all. The analyst-level pain is acute and largely unaddressed by AI tooling. The buyers are smaller, more willing to experiment, and not already locked into enterprise AI contracts for this specific workflow. The switching cost is low because there’s nothing to switch from.

Same problem type. Different market to build in.

The lesson isn’t that AI for legal workflows is a bad idea — it clearly works and there’s real demand. The lesson is that “the problem exists and AI can solve it” is insufficient analysis. You have to understand who already owns the customer relationship, what their switching costs look like, and whether you’re entering a workflow gap or competing against an entrenched incumbent for a buyer who has already made a choice.

Choosing the wrong battleground is how technically sound products fail in otherwise good markets. The AI capability problem is often the easier one to solve. The market access problem is the one that actually determines outcomes.

+++