Bodify

Bodify is your multi-agent AI health partner. A council of AI specialists creates personalized nutrition, fitness, and lifestyle plans that adapt to your goals, health, and life's setbacks.

Comment

Inspiration

In Indonesia, only around 65% of Puskesmas meet minimum staffing requirements of nine basic health worker categories (lacking of doctors, nurses, nutritionists, etc), and the country is projected to face a shortage of around 65,000 medical specialists by 2032 (WHO, 2026). Even when patients reach a doctor, consultations are often too brief to support the daily decisions that chronic conditions like diabetes, hypertension, and obesity require. Meanwhile, generic health advice ignores the reality that every body is different (age, body composition, medical conditions, lifestyle, goals, or even psychological barriers!). That's why I created Bodify to make continuous, personalized health guidance accessible to everyone, anywhere, without exception.

What it does

Bodify is a multi-agent AI health partner. A council of specialist agents (medical, nutrition, fitness, behavioral, and progress), reviews the user's profile, goals, daily logs, and life events. A coordinator agent (Bodi) resolves conflicts and produces a personalized plan that adapts to user's daily life activities. Through the dashboard, users can:

  • Home — See today's schedule, streak, progress metrics, and log their day.
  • Calendar — View the adaptive calendar managed by the Agent Councils.
  • Recovery — Report setbacks like travel, injury, illness, or social events and get an adapted plan.
  • Simulate — Run "what-if" scenarios (e.g., missing workouts, changing diet) and see projected impact.

Users can also speak instead of type (Speech-to-Text technology). A microphone button in the daily log, recovery, and simulate pages records audio, making it easier for people who's tired throughout the day.

How I built it

I built Bodify using:

  • Frontend: Next.js, React, TypeScript, Tailwind CSS, shadcn/ui, Framer Motion, GSAP, Recharts, Valibot
  • Backend: FastAPI, Uvicorn, Pydantic, SQLAlchemy, Alembic
  • Relational Database and Auth: Supabase
  • Vector Database: ChromaDB
  • RAG: Medical guidelines for diabetes, hypertension, exercise, and behavior (from NIH, ADA, WHO, NHLBI)
  • AI / LLM: LangChain, LangGraph, OpenRouter (Meta Llama 3.3 70B Instruct for LLM)
  • Multi-Agent System: Specialist agents (medical, nutrition, fitness, behavioral, progress), orchestrator agent, coordinator agent, fact extractor
  • Speech-to-Text: Deepgram Nova-3
  • AI Tools Used: ChatGPT (generated logo and mascot), OpenCode (GLM 5.2), 9Router (RTK on throughout the whole development, Caveman and Ponytail for building features)

Challenges I ran into

Lack of sleep, and failed on Vercel deployment (__dirname is not defined issue, failed to debug it), even though Railway deployment succeeded (https://bodify-production.up.railway.app/health).

Accomplishments that I'm proud of

I'm proud of having this project done.

What I learned

Time management is surely the most important thing.

What's next for Bodify

What I have in my mind:

  • Integrate Google Calendar to have more personalized Calendar
  • Integrate with wearable devices (e.g Garmin, Google Fit, Apple Watch) to gather more data.
  • For meal logging, integrate Computer Vision.
  • Add budgeting even though its hard to implement since every place has different prices.
  • Continue to build this project for a better world.

Built With

  • 9router
  • chatgpt
  • chromadb
  • framer-motion
  • gsap
  • langchain
  • langgraph
  • next.js
  • opencode
  • openrouter
  • pydantic
  • rag
  • react
  • recharts
  • shadcn/ui
  • stt
  • supabase
  • tailwind-css
  • typescript
Share this project:

Updates