Codes by Shrey

Practice Intuition

Bonita Encode

An AI wellness intelligence MVP that helps users route real-world challenges into TIME, SPACE, and SELF, then receive one structured protocol with clear steps, rationale, and safety boundaries.

Product System

  • TIME: circadian rhythm, sleep, timing, energy
  • SPACE: body, movement, pain, fascia, environment
  • SELF: mood, cognition, emotional regulation, identity
  • IQ / EQ / KQ wellness dashboard framing

Domain

AI wellness intelligence

Stack

React, TypeScript, Gemini

Pattern

Assessment to protocol

Safety

Medical boundaries and escalation

Problem

Most wellness products isolate sleep, exercise, mood, or tracking. Bonita treats wellness as an interconnected system and gives users one actionable next step instead of an overwhelming plan.

Experience

Users can start from a branded entry point, browse pillar resources, complete a chat-based assessment, add daily check-in context, and generate a broader wellness dashboard.

AI Behavior

The assistant classifies user needs, asks focused clarifying questions, returns structured JSON, renders checklist-style protocols, and preserves clear boundaries around medical advice.

Product Value

The MVP demonstrates PRD-to-product translation: a premium-feeling front end, structured LLM output, safety checks, local telemetry concepts, and an extensible protocol framework.

Process + Progress

The product is documented as a PRD-to-MVP translation: a structured assessment flow, protocol engine, safety boundary model, and future telemetry loop.

PRD

Define the TIME, SPACE, SELF pillars and the wellness intelligence framing.

Embedded below

MVP

Build a polished entry point with assessment, pillar resources, and protocol output.

Live build linked from hero.

Protocol

Use structured LLM output so the assistant returns renderable product data instead of raw chat.

Checklist-style protocol workflow.

Iteration

Next layer: saved progress, protocol quality review, telemetry, and follow-up loops.

Roadmap-ready product surface.

Technical Skills Demonstrated

Structured LLM Output

Classification, clarifying questions, JSON-style responses, and UI-rendered protocols.

Product Safety

Medical boundary setting, escalation language, and wellness-coaching constraints.

Frontend Delivery

React, TypeScript, branded interaction design, and PRD-to-product implementation.

System design / architecture details
Input model

Users describe a current wellness challenge in natural language, optionally adding daily check-in context before protocol generation.

Routing model

The assistant classifies the challenge across TIME, SPACE, SELF, or multi-pillar patterns so the response has a clear conceptual home.

Output model

The model returns structured coaching data that can render as checklist-style protocol cards rather than loose chat advice.

Bonita was the proof-of-concept layer for a broader Somanaut direction: a coaching system where wellness intelligence is structured, safety-aware, and repeatable enough to become a product rather than a conversation demo.

Bonita Encode PRD details

Product Summary

Bonita Encode is a consumer-facing MVP for integrated wellness guidance. It helps users name a current problem, classify the issue across TIME, SPACE, and SELF, and receive one protocol that is specific enough to try immediately.

Problem

Many wellness tools are too narrow, too generic, or too dependent on passive tracking. Users need personalized guidance that connects body, mind, timing, and environment while teaching the reasoning behind each recommendation.

Core User Stories

  • Describe a current wellness challenge in natural language.
  • Understand whether the challenge is primarily TIME, SPACE, SELF, or multi-pillar.
  • Receive one practical protocol with steps and time cost.
  • Learn why the protocol works.
  • Generate a broader wellness plan after enough conversation.
  • Receive clear escalation when an issue may require medical help.

Functional Requirements

  • Allow users to start from a home screen, browse resources, and drill into a pillar before chat.
  • Infer likely pillar routing from user messages.
  • Render structured JSON coaching responses as checklist-style protocol cards.
  • Pass optional daily check-in context into the next AI request.
  • Run deterministic red-flag checks before model generation.
  • Generate a structured wellness dashboard with IQ, EQ, KQ scores and recommendations.

Risks and Gaps

Future versions should move model calls behind a secure backend, add persistence, expand safety coverage, support citations or evidence provenance, and turn mock longitudinal concepts into real progress tracking.

Product Vision

Bonita can evolve into a premium AI wellness intelligence platform that combines personalized behavior-change support, integrative health education, and longitudinal self-awareness while staying clear about medical boundaries.