Track 1 : Agentic AI
Team Name : BotArmy
प्रारम्भ means The Commencement in Sanskrit. That is exactly what this is — the first thing a new engineer touches when they join.
Most onboarding is broken. Engineers spend their first week sending Slack DMs, waiting for access, reading outdated wikis, and asking "what do I do next?" Prarambh fixes that with a multi-agent AI system that guides every new hire through the entire process autonomously.
A new engineer opens the platform, introduces themselves, and the system takes it from there.
It figures out who they are, loads a personalized checklist based on their role and experience, walks them through each task, answers questions from the company knowledge base, tracks completion, provisions tool access, and sends an HR report when they're done — all without a human in the loop.
Seven specialized agents, each with a single responsibility. Guru receives every message and decides which agent should handle it.
| Agent | Meaning | What it does |
|---|---|---|
| Guru (गुरु) | Guide | Root orchestrator — reads intent, routes to the right agent |
| Parichay (परिचय) | Introduction | Identifies the employee — role, level, tech stack |
| Dwar (द्वार) | Gateway | Provisions GitHub, Slack, Jira, Notion, AWS access |
| Prakriya (प्रक्रिया) | Process | Drives the step-by-step onboarding checklist |
| Gyaan (ज्ञान) | Knowledge | Answers questions using RAG over internal docs |
| Karma (कर्म) | Action | Marks tasks complete, updates progress, assigns tickets |
| Sandesh (सन्देश) | Message | Generates and sends the HR completion report |
These are not one LLM pretending to be seven — each is a separate LlmAgent instance in Google ADK with its own instructions and tools.
| LLM | Gemini gemini-3.1-flash-lite-preview |
| Agent Framework | Google Agent Development Kit (ADK) |
| Backend | FastAPI + Uvicorn |
| RAG / Vector DB | ChromaDB |
| Frontend | Next.js 14 (App Router) + TypeScript |
| Styling | Tailwind CSS + shadcn/ui |
| Animation | Framer Motion |
| State | Zustand |
| Real-time | WebSocket |
prarambhai/
├── backend/
│ ├── agents/gurukul.py — 7 ADK agents
│ ├── tools/agent_tools.py — all agent tools
│ ├── rag/ — ChromaDB ingestion + retrieval
│ ├── models/ — Pydantic data models
│ ├── knowledge-base/ — internal docs (RAG-ingested)
│ ├── templates/ — HR email templates
│ └── main.py — FastAPI server + WebSocket
│
└── frontend-next/
├── app/ — Next.js App Router
├── components/ — Sidebar, TaskPanel, AssistantPanel
├── hooks/use-websocket.ts — WebSocket with auto-reconnect
├── lib/store.ts — Zustand global state
└── types/ — TypeScript types
Prerequisites: Python 3.11+, Node.js 18+, Google Gemini API key
python -m venv venv
venv\Scripts\activate # Windows
# source venv/bin/activate # macOS / Linux
pip install -r backend/requirements.txt
copy .env.example .env
# Add your GOOGLE_API_KEY to .env
cd backend
python -m uvicorn main:app --host 0.0.0.0 --port 8000cd frontend-next
cp .env.local.example .env.local
npm install
npm run devOpen http://localhost:3001 and introduce yourself to get started.
User: "Hi, I'm Arjun. Joining as a Backend Engineer, 2 years experience, Python."
Guru: Parichay has identified you as a Backend Engineer (Mid-level, Python).
Dwar is provisioning your GitHub, Slack, and Jira access now.
Prakriya has loaded your 34-task onboarding checklist. Task 1: Set up local dev environment.
User: "Done."
Guru: Karma has marked Task 1 complete. Progress: 3%.
Next: Configure your VPN access. Here is the setup guide...
Seven internal documents are ingested into ChromaDB at startup:
- Company overview and org structure
- Engineering standards and coding guidelines
- Architecture documentation
- Tool setup guides
- HR policies
- Onboarding FAQ
The Gyaan agent retrieves relevant chunks using semantic search when an employee asks a question.
Built for Syrus 2026 by Team BotArmy