Why Voice AI for Medical Records Has to Be Boring
Building MedVoq taught me that the most important AI features are usually the ones nobody notices.
The Job Is Not to Sound Smart
MedVoq listens to a live doctor and patient consultation and turns it into a structured, doctor-approved medical record. It would be easy to build something that feels impressive in a demo, fluent, fast, conversational, and still be wrong in ways that matter.
The actual job is narrower and less glamorous: extract medications, dosages, and instructions into a strict schema, every time, even when the audio is messy.
Designing for Messy Input
Real consultations are accented, multilingual, and full of the kind of code switching that breaks a lot of transcription pipelines. Instead of chasing a smarter model, I constrained the output to a strict schema and built verification around it, so the system fails loudly instead of quietly.
A Human Stays in Control
Every record MedVoq produces goes through a doctor-in-the-loop approval step before it becomes final. That is not a limitation I am working around, it is the actual safety mechanism. The AI's job is to remove the tedious part of documentation, not to make the final call.
More Articles

Feb 10, 2026
Making Crypto Feel Invisible to the User
Building LazorTicket meant hiding wallets, gas fees, and NFTs behind something as simple as Face ID.

Mar 3, 2026
Building a No-Code Trading Platform From Scratch
How Quanto lets retail traders build and backtest strategies without writing a line of code.

Apr 21, 2026
Why AI Agents Need a Human in the Loop
Lessons from building the agentic marketing system behind Klyq AI, where every action needs a yes or no from a person.