There are two distinct ways to build in a market that’s fragmenting into vertical specialists.

The first is to become another specialist. Pick a workstream, go deep, build the best tool for that slice. This is the strategy that produces the agentic accounting company, the environmental due diligence platform, the lease abstraction tool. Each of these is a focused bet on one slice of the workflow, built for the professionals who care most about that slice.

The second is to build at the protocol layer. Rather than owning a workstream, you own the integration. You connect the specialist tools through a standard protocol, unify the workflow for the end user, and let the specialists handle their domains. You’re not competing with the verticals — you’re the layer that orchestrates them.

These two positions compete differently. A specialist competes with other specialists in the same workstream. A protocol-native tool competes with the workflow itself — specifically, with the manual stitching that professionals currently do to move information between specialists.

The protocol-native position has an interesting property: it gets stronger as the specialist ecosystem matures. Every new specialist that ships with MCP support is potential input to the protocol-native layer. The specialist’s success validates demand; the protocol-native tool captures the integration value.

The risk is timing. Build the integration layer before the specialists exist and you have nothing to integrate. Build it after the specialists have already shipped their own integrations and you’ve missed the window.

The right timing is when the specialists have proven the market but haven’t solved the stitching problem. That’s the window. +++