AI Voice Depends on Workflow
Fonio.ai
The real moat in vertical AI voice is not the model, it is control of the customer workflow inside the system of record. Fonio.ai can answer a call with commodity speech and language tools, but high value actions like finding an open appointment, booking the right slot, pulling patient or job history, and writing the result back into the record depend on deep access to the software that runs the business. Doctolib is building a healthcare operating system with a virtual phone assistant, and ServiceTitan now sells its own tightly integrated phone stack, which makes that control especially strategic.
-
Doctolib is no longer just a calendar. It presents itself as a full operating system for health professionals, spanning scheduling, patient messaging, intake, financial tools, and digital secretariat products including Phone Assistant and Virtual phone assistant. If voice becomes one more native module in that stack, an outside assistant risks looking shallow by comparison.
-
ServiceTitan shows the field service version of the same pattern. Its Phones Pro product ties inbound calls directly to job booking, call attribution, and CSR performance tracking inside ServiceTitan. Its API program also gives the company formal levers over who can distribute apps, what scopes they get, and whether AI based applications face extra conditions or write restrictions.
-
This is why workflow depth matters more than call quality. A standalone assistant that can only answer common questions is easy to replace. One that can check availability, book the right technician or clinician, attach notes to the customer record, and trigger downstream workflows can charge more and convert better, but only if the record system allows it.
Going forward, the winners in AI voice will be the products that either own the record system or become indispensable distribution partners to it. For Fonio.ai, expansion into verticals will increasingly hinge on securing privileged integrations before platforms like Doctolib and ServiceTitan turn voice into a native feature and narrow the room for standalone layers.