FounderOS — AI Chief of Staff

Voice-first execution partner and semantic memory hub for solo founders.

Hackathon Submission for HiDevs "Dawn of the Autonomous AI Builder"
Theme: "One Man, An Entire System" — A voice-first digital workspace converting spoken thoughts into structured decisions, tasks, and pattern insights.

🌐 Live Production Links

• Frontend App (Vercel): https://founder-os-pi-lemon.vercel.app/
• Backend Service (Render): https://founderos-backend-uuuv.onrender.com/

FounderOS captures everything a founder says via voice (integrated with Omi), processes transcripts through specialized AI agents (Lyzr), stores them in a cloud semantic memory (Qdrant), and lets them query their execution history conversationally.

---

📸 Dashboard Preview

FounderOS features a dark-mode dashboard tailored for visual excellence and high performance, featuring:

• Voice Q&A: Web Speech API voice capture + text-based semantic retrieval.
• Activity Timeline: Filterable feed of Decisions, Tasks, and Ideas with direct status toggles.
• Pattern Insights: Staggered, synthesized breakdowns of recurring themes and work blockages.
• Privacy Controls: Complete control over your semantic database, including granular memory resetting.

---

🏗️ Architecture

 ┌──────────────┐       ┌───────────────────────────────────────────┐
 │  Omi Device  │──────▶│           FastAPI Backend                 │
 │  (Voice In)  │  POST │                                           │
 │  /api/omi/   │  /api │  ┌─────────────────────────────────────┐  │
 │  webhook     │  /omi │  │    Lyzr Agent Pipeline (Parallel)    │  │
 └──────────────┘  /hook │  │  ┌────────────┐ ┌──────────────┐    │  │
                         │  │  │  Decision   │ │    Task      │    │  │
                         │  │  │  Extractor  │ │  Extractor   │    │  │
                         │  │  └─────┬──────┘ └──────┬───────┘    │  │
                         │  │  ┌─────┴──────┐        │            │  │
                         │  │  │    Idea     │        │            │  │
                         │  │  │  Extractor  │        │            │  │
                         │  │  └─────┬──────┘        │            │  │
                         │  └────────┼───────────────┼────────────┘  │
                         │           ▼               ▼               │
                         │  ┌─────────────────────────────────────┐  │
                         │  │      Qdrant Vector Database          │  │
                         │  │  decisions│tasks│ideas│patterns      │  │
                         │  └─────────────────────────────────────┘  │
                         └───────────────────────────────────────────┘
                                           ▲
                                           │  REST API (with CORS)
                         ┌─────────────────┴──────────────────────┐
                         │        React Dashboard (Vite)          │
                         │  Voice Q&A │ Timeline │ Patterns │ Privacy │
                         └────────────────────────────────────────┘

---

🛠️ Technology Stack

Layer	Technology	Purpose
Voice Capture	Omi	Captures conversational audio & streams completion webhook
Agent Orchestration	Lyzr Core API	Orchestrates 4 parallel agents extracting insights from transcripts
Vector Memory	Qdrant Cloud	High-performance semantic vector search and metadata retrieval
Embeddings	OpenAI	Generates 1536-dimensional embeddings (text-embedding-3-small)
Backend API	FastAPI	REST API routing, webhook parsing, and CORS control
Frontend UI	React (v19) + Vite	SPA with responsive typography, dark theme, and SVG micro-interactions
Guardrails	Enkrypt AI	Inspects outputs for PII and alignment safety

---

⚙️ Environment Configuration

Create a .env file in the project root containing your secret keys. Both local runs and production web hosts rely on these values:

# ── LYZR AGENTS CONFIG ────────────────────────────────────────────────────────
LYZR_API_KEY=your_lyzr_api_key

# ── OPENAI CONFIG ─────────────────────────────────────────────────────────────
OPENAI_API_KEY=your_openai_api_key

# ── QDRANT VECTOR STORAGE CONFIG ──────────────────────────────────────────────
# Local development default: http://localhost:6333
# Production cloud cluster: https://<cluster-id>.<region>.aws.cloud.qdrant.io
QDRANT_URL=https://your-qdrant-cluster-url.aws.cloud.qdrant.io
QDRANT_API_KEY=your_qdrant_api_key

# ── ENKRYPT AI INTEGRATION (OPTIONAL) ─────────────────────────────────────────
ENKRYPT_API_KEY=your_enkrypt_api_key

# ── BACKEND SECURITY ──────────────────────────────────────────────────────────
# Comma-separated list of allowed CORS origins
CORS_ORIGINS=http://localhost:5173,https://founder-os-pi-lemon.vercel.app

---

💻 Local Development

1. Run via Docker Compose (Easiest)

Build and run all services (Qdrant, Backend, and Frontend) in background containers:

docker-compose up -d --build

• Frontend: http://localhost:5173
• Backend Docs: http://localhost:8000/docs
• Qdrant Dashboard: http://localhost:6333/dashboard

2. Manual Development Run

If you wish to run the processes natively:

Start the FastAPI Backend

cd backend
python -m venv .venv
source .venv/bin/activate  # Or `.venv\Scripts\activate` on Windows
pip install -r requirements.txt
uvicorn app.main:app --reload --port 8000

Start the Vite React Frontend

Open a separate terminal window:

cd frontend
npm install
npm run dev

Access the application at http://localhost:5173.

---

🚀 Production Deployment Guide

Follow these steps to deploy FounderOS to cloud environments.

1. Backend Web Service (Render)

Render is ideal for hosting the FastAPI backend web service.

1. Create Web Service: Link your GitHub repository in Render and create a new Web Service.
2. Configure Base Directory: Set the root directory of the build to backend.
3. Build & Start Commands:
   • Runtime: Python
   • Build Command: pip install -r requirements.txt
   • Start Command: uvicorn app.main:app --host 0.0.0.0 --port $PORT (If deploying from the root workspace, configure the build command as pip install -r backend/requirements.txt and start command as uvicorn backend.app.main:app --host 0.0.0.0 --port $PORT)
4. Environment Variables: Add all environment variables listed in the configuration section (e.g. LYZR_API_KEY, OPENAI_API_KEY, QDRANT_URL, QDRANT_API_KEY, and CORS_ORIGINS). Make sure CORS_ORIGINS includes your production Vercel frontend URL.
5. Verify Endpoint: Check Render's health endpoint: https://your-service.onrender.com/api/health.

2. Frontend SPA (Vercel)

Vercel is perfect for serving the Vite React frontend.

1. Create Project: Import your GitHub repository in Vercel.
2. Configure Folder: Set the Root Directory of the project to frontend.
3. Build & Output settings: Vercel automatically detects the Vite template.
   • Framework Preset: Vite
   • Build Command: npm run build
   • Output Directory: dist
4. Environment Variables: Add the following variable to configure routing:
   • VITE_API_URL: Set to your production Render URL (e.g., https://founderos-backend-uuuv.onrender.com).
5. SPA Redirect Routing: Vercel utilizes the local vercel.json configuration inside the frontend folder to rewrite all client paths to index.html, ensuring clean browser history navigation.

---

🎙️ Hooking Up Omi Device (Voice webhook)

For real-time voice logging via the Omi audio device:

1. Expose Localhost (Development): Expose your local backend using ngrok:

   ngrok http 8000

2. Register URL (Developer Portal): In the Omi developer dashboard, register your webhook endpoint:
   • Webhook URL: https://your-subdomain.ngrok-free.app/api/omi/webhook (or your production Render url: https://your-backend.onrender.com/api/omi/webhook)
   • Subscribe Events: conversation.completed
3. Action Flow: When you speak and finish a conversation, Omi sends the audio transcript to the webhook. FounderOS extracts decisions/tasks/ideas in the background, updating your dashboard timeline automatically.

---

📖 API Reference

Method	Endpoint	Description
POST	/api/omi/webhook	Receives Omi transcripts, runs parallel agents, and saves results to Qdrant
POST	/api/query	Queries semantic memory with voice/text inputs
GET	/api/timeline	Browses all Decisions, Tasks, Ideas, and Patterns with filters
POST	/api/digest	Generates weekly pattern digest analysis of recent logs
GET	/api/collections	Counts entries across all active Qdrant databases
PUT	/api/entries/{collection}/{id}	Updates metadata of a specific entry (e.g. task done status)
DELETE	/api/entries/{collection}/{id}	Deletes a specific entry from Qdrant memory
DELETE	/api/collections/{name}	Clears all points in a collection (Resets database)
GET	/api/health	Service health status check

---

⚖️ License

Developed for the HiDevs "Dawn of the Autonomous AI Builder" Hackathon. All rights reserved.