ποΈ High-Fidelity Clinical Interface for Neurophenomenology
Live demo: neurophenomai.newpsychonaut.com
κ© Map pre-reflective subjective experience through granular interview techniques.
ποΈ Use the trained AI research interviewer or record and transcribe.
𧩠AI analyses and codifies each interview.
High-fidelity clinical interface for mapping pre-reflective subjective experience. The conversational AI is trained in granular microphenomenology interview techniques, on altered states of consciousness in particular. Codify and theme your subject data alongside NeuroPhenom AI's granular LLM engine. Analyse the micro-dynamics of lived moments through diachronic slicing and structural synthesis.
Featuring a stark black-and-white minimalist design, the interface stays out of the way and lets the phenomenological work take centre stage.
- Node.js 22 (Volta-pinned in
package.json) - Google Gemini API key (entered in the in-app Settings menu; stored in the browser session)
- Chrome recommended for live interview audio (Web Audio + Gemini Live). Select your USB microphone in Settings; prefer wired headphones during AI interviews.
- Cloud Run demo clone (exact AI Studio source): sibling repo NeuroPhenomAI-Showcase β see docs/HANDOFF_2026-07-10.md. Gold applet:
https://neurophenom-ai-572556903588.us-west1.run.app/. In-repo Cloud Run notes: docs/CLOUD_RUN_SHOWCASE.md.
- ποΈ Live Interview Sessions β AI-guided microphenomenology interviews in real time
- π€ Standalone Recorder β Capture audio independently of the interview system
- π Analysis View β Review and codify interview data with AI assistance
- π¬ Diachronic Slicing β Temporal decomposition of experiential moments
- 𧩠Structural Synthesis β Map the architecture of subjective states
- β Consent Protocols β Built-in consent management for ethical research practice
- πΎ Local Storage β Interviews saved locally for data sovereignty
- βοΈ Configurable β Language, microphone device, AI voice, and interview mode settings
- π€ Minimalist Design β Clean black/white aesthetic focused on the work
| Layer | Technology |
|---|---|
| Frontend | React 19 + TypeScript, Vite |
| Client AI | Google Gemini API (BYOK via Settings) |
| Production server | Express (server.js) serving dist/ |
| Optional server AI | Azure OpenAI / Foundry (/api/* β not used by the UI today) |
| Deployment | Azure App Service (primary); Cloud Run showcase (Gemini proxy) |
index.html β index.tsx β App.tsx
ββ LiveInterviewSession ββ
ββ AnalysisView βββ services/geminiService (browser Gemini)
ββ StandaloneRecorder ββ
npm start β server.js β static dist/ + /api/health|/api/welcome|/api/analyze (Azure Foundry)
- Today: the React app talks to Gemini in the browser with your own API key (privacy-first BYOK).
- Server Foundry routes in
server.jsare ready for a future server-mediated path; the UI does not call them yet. - Methodology baseline:
docs/knowledge/core/NP_CANONICAL_SPEC.md - Public one-pager:
docs/github-pages/index.html
git clone https://github.com/chaosste/NeuroPhenomAI.git
cd NeuroPhenomAI
npm install
npm run dev- Open the app (Vite defaults to
http://localhost:8080). - Open Settings and paste your Google Gemini API key.
- Start a live interview or use the standalone recorder.
Production-style local run (build + Express):
npm run build
npm startServer env vars (optional Azure Foundry, rate limits) are documented in .env.example.
Use the worktree/branch protocol in:
Operational canon:
π‘ Like NeuroPhenom AI? You'll love MicroPhenom AI β the vanilla edition for granular reports on wider lived experience.
π For psychedelic trip report interviews with theatrical deity-themed voices, see Anubis.
NeuroPhenom AI is a research tool for exploring subjective experience. It does not provide medical, psychological, or therapeutic advice. It is not a substitute for professional support.
