Shannu.
← Back

03 / The Work

Satya

App Store · Play Store, Approved First Attempt · Full Feature Set Live · Paid Tier

Built At Antz →

The Problem

Building a medical-adjacent AI product that needed to earn trust before it earned features — get the core assistant right, get it approved, get it in front of real users, before adding anything that raises the stakes.

The Decisions

Approved On The First Try

Play Store approval landed on the first submission, which doesn't happen by accident in a regulated-adjacent category.

Sequenced The Build Deliberately

Core assistant first, voice second, once the base product had proven itself with real users. Trust before capability, in that order.

The Handoff Was Part Of The Deliverable

Built it to be run by someone else, not maintained indefinitely by me.

How It Works

  1. 01Open the app, ask a health question, get an answer built to be careful before it's clever — the whole point in a medical-adjacent product.
  2. 02Prefer to talk? Voice in, voice out, added only once the text assistant had earned its place with real users.
  3. 03It's a paid product now, on both stores, and it isn't mine to babysit — the team that runs it day to day took the keys.

High-Level Design

Mobile App (RN)API (FastAPI)LLM OrchestrationModel ProviderContext StoreRemindersguardrails · persona· doc contextconditions · meds · history
// sanitized · generic pattern, no patient data or real service names

The Outcome

Live on both stores, full feature set including voice, now in its paid tier and handed off to the team running it day to day.

live on both stores · marketing not started yet, so there are no usage numbers to quote

Shipped It, Then Made Myself Redundant. On Purpose.