Build Journal · AyurDoc
Building AI that suggests the next question without leading the consult
By Ganesh R Datta. I build healthcare AI end to end: clinical decision support, clinic intake, and the marketing and operations around them.
AyurDoc surfaces matching prior visits and suggests the next question to ask. The design rule is that suggestions stay optional and visible as suggestions, so the practitioner leads the consult and the software helps them not miss things.
There is a fine line between helping a practitioner and steering them. Clinical decision support has to stay on the helping side.
What the feature does
During the consult, AyurDoc can surface similar prior visits and suggest a question the practitioner might ask next. It is a nudge toward completeness: the cause not yet explored, the history not yet taken.
Rules I hold it to
- Suggestions look like suggestions. They never appear as conclusions.
- Ignoring a suggestion costs nothing. No extra clicks to dismiss.
- The practitioner's own line of questioning always takes priority over the software's.
Why this matters
Practitioners are experts with their own judgement. A tool that tries to lead them gets switched off. A tool that quietly catches the question they would have asked on a less busy day earns a place in the room.
The broader principle
In every CDSS I build, the software prepares and the clinician decides. Prompting is part of preparing. Diagnosing is not.
More from the Build Journal
The full Build Journal · The AyurDoc work · Book a 15-minute call