# inou health — Context for James

*Updated: 2026-02-20. Always lowercase "inou" — avoids L/I confusion.*

## What it is
**Medical data storage platform / infrastructure.** Born from Sophia's accident (May 2, 2022). Johan built it to fight the medical system — read MRIs, challenge diagnoses, build evidence for specialists. It's not a wellness app, and it's not an AI service. It's infrastructure for people the system has failed.

**⚠️ CRITICAL FRAMING:** inou = storage platform, infrastructure layer. AI is pluggable — not the product. The goal is to support ALL major LLMs: Claude, ChatGPT, Grok, Kimi, MiniMax, you name it. inou holds your medical data; you connect whatever AI you want to it.

**Strategic direction (2026-03-13):** Build connectors/integrations for every major model/platform. MCP is the Claude connector — replicate this pattern for ChatGPT (plugin/GPT connector), Grok, MiniMax, Kimi, etc. Data schema must be clean and well-structured so any AI gets good results.

Nudges should focus on: new LLM integrations, connector reliability, data schema quality, storage infrastructure, import pipelines (DICOM/FHIR/lab), data portability. NOT: "build an AI feature."

## Current capabilities
- **DICOM viewer + analysis** — brain MRIs, CT scans, radiology reports
- **Genetic analysis** — SNPedia integration, variant interpretation
- **Lab data import** — MyChart integration, manual upload
- **Mobile health logging** — vitals, medications, observations (API)
- **Claude MCP integration** — Claude can query inou data directly
- **6-language support** — including Dutch
- **Sophia-specific tooling** — the original use case drives design decisions

## Business model
- **Monitor** (free) — basic access
- **Optimize** ($12/mo) — fuller feature set
- **Research** ($35/mo) — deep analysis tools
- **Free until July 1, 2026** — early access period, no active promotion yet

## Status: building phase
Not ready to promote. Johan's directive: focus on product, not word-of-mouth.
Do NOT suggest marketing, tweeting, press, or going public with suggestions.

## Where the code lives
- `/home/johan/dev/inou` — Johan uploads portions via SMB share (`inou-dev`)
- "Nibble" approach — James works on what Johan shares

## Known gaps / areas to explore
- Citation/reasoning transparency (Superpower does this well — show Claude's reasoning chain tied to sources)
- Better onboarding for non-technical users (doctors, patients)
- Wearable data integration (Superpower roadmap item — worth considering)
- FHIR/HL7 compatibility for importing records from other systems
- Sophia case: MRI analysis pipeline needs to stay current with new scans
- Mobile app (inou-mobile in Flutter) — thin shell + WebView, still early

## Key people
- **Dr. Neel Madan** (Chief Neuroradiology, Tufts) — reviewing Sophia's MRI → next step: neurosurgery
- **Johan** — the only user who matters right now. Every feature should serve his actual workflow.

## Design principles
- Evidence over opinion — everything citable
- Local-first for sensitive data (medical docs never hit external APIs)
- Built for advocacy, not wellness optimization
- Johan's night-shift constraint: usable at 2am, one-handed, while monitoring Sophia