// REFERENCE PROJECT: HEALTH-TECH · DNA & HUMAN-OPTIMIZATION APP

Identity Code

A DNA coaching process that used to run on ChatGPT uploads, rebuilt as an app of its own.

Identity Code used to be a DNA coaching process that ran on PDFs uploaded into ChatGPT. We turned it into a web app of its own as a product base. Eight configured expert profiles are designed to guide members by chat based on their DNA report; in the handed-over demo build this chat ran on an LLM mock. Members, practitioners and admins each get their own surface. Challenges, structured check-ins, DNA report assignment and an abstraction for the AI provider come on top, so the platform is not tied to a single vendor.

Open app access →

IDENTITY CODE

Digitale Experten

Psyche

Resilienz

Fitness

🔒Ernährung

🔒Health

🔒Fertilität

Product illustration based on the original German interface.

Industry

Health-tech · DNA coaching

Type

Web app / PWA with member, practitioner and admin surfaces

Cycle

Three phases: foundation, challenge mode, practitioner and handover

Stack

React · Vite · TypeScript · Node/Express · SQLite · lucide-react

// THE STARTING POINT

Kevin wanted to move Identity Code from a visible ChatGPT link to a product of its own: a black-and-gold world of DNA and AI that feels modern yet stays simple enough for an audience partly over 50. The combination of DNA, epigenetics, neurobiology and AI had to become concrete routines without ever claiming medical advice. The reference material was strong but overloaded: too many panels, too much glow, text too small. We kept the mood, cut the information density hard and built the member, practitioner and admin flows including cost, DNA and entitlement logic.

// WHAT WE BUILT

✓Black-and-gold design system: Cinzel headlines, Montserrat body text, gold as accent rather than fill, glow only for active states

✓Member app: calm dashboard, expert picker with locked and unlocked experts, expert chat with context (DNA basis, focus areas, recommendations)

✓Challenge mode with structured check-ins (energy, mood, sleep, weight, stress) and progress and goal tracking

✓Practitioner surface: client list, active client context, experts and chat per client, notes and insights

✓Admin: member management, invites and password reset, DNA report assignment, cost and token limits and knowledge sources

✓AI provider abstraction (mock plus prepared OpenAI and Claude adapters) with a guard for medical boundaries and server-side usage logging

// HOW THE PROJECT RAN

Phase 1: Goal and scope

From vision to a usable baseline

From the call, website and mockups we distilled the core: digital experts as the product, chat as the main interaction, black and gold as the brand world. The guiding decision was not to rebuild the overloaded designs but to translate them into a calmer interface. The three roles and the firm “no medical advice” boundary framed the scope.

Phase 2: Prototype

Design system and expert logic, clickable

The design tokens were set: near-black surfaces, a gold accent, Cinzel and Montserrat, glow only for active states. A first clickable build with login, expert picker and expert chat with context was enough to test the reduced information density against the references.

Phase 3: Build

Three surfaces, challenges and cost control

We built out the member, practitioner and admin flows: challenge mode with check-ins, progress and goals, client management per practitioner and an admin with DNA assignment, entitlements and cost and token limits. The AI provider is abstracted (mock plus prepared adapters), with a medical-boundary guard and server-side usage logging.

Phase 4: Handover

Demo-ready and cleanly handed over

We handed over a single-server demo build with real login (bcrypt, HttpOnly sessions), a SQLite data model and a smoke test across all roles, the LLM mock and the safety guard, together with a handover document, an acceptance checklist and documented next steps for the follow-on developer.

Outcome

role surfaces (member, practitioner, admin) from one codebase

configured expert profiles, each with its own briefing and unlock logic

Status and limits

At our handover the chat experts ran on an LLM mock; the OpenAI and Claude adapters were prepared, and the production connection was the next step for the follow-on developer. We make no statement about the current state of the AI connection. The link opens the app’s login. The app deliberately gives no medical advice.

NEXT PROJECT

Viking Hands →

Does your offer run on documents and a general-purpose AI chat today? Describe your plan and get a first assessment of scope and timeline.