UniGen is an AI agent for on-chain intelligence. You ask in natural language; UniGen fetches wallet, token, and transaction signals, then explains them — not just dumps charts.

// UniGen_

What is UniGen?
→ AI agent for on-chain intelligence

What can I do?
→ Analyze wallets, tokens & transactions

Where?
→ Robinhood

1. Open /app
2. Type a query or pick a suggestion chip
3. Watch [ Processing... ] → structured panel + AI narrative

$ unigen_
> Analyze this wallet
> 0x7A...92F

[ Processing... ]
> indexing on-chain data...
> ready_

The terminal is the product surface. Language matches the visual system:

• // comments
• <components /> tags
• $ commands
• [ status ] state
• → data flow
• _ cursor

Input accepts free-form English. UniGen classifies intent (wallet / token / transactions / general) and returns matching intelligence.

Differentiation is interpretation. Raw fields sit in the dashboard; the AI layer answers what it means.

UniGen AI:

This wallet has increased ETH exposure
by 34% over the last 7 days.

Recent activity suggests
accumulation rather than selling.

MVP uses curated mock intelligence for demo reliability. With XAI_API_KEY set, /api/analyze can enrich narratives via SpaceXAI / xAI.

WALLET
├── Balance
├── Transactions
├── Holdings
├── PnL
└── Activity

TOKEN
├── Price
├── Volume
├── Holders
├── Liquidity
└── On-chain Activity

Phase 1 keeps charts minimal on purpose. Structure + narrative first; deep viz later.

Client and server share one endpoint for the terminal agent loop.

POST /api/analyze
Content-Type: application/json

{ "query": "Analyze this wallet 0x7A..." }

→ {
  kind, statusLines, insight,
  wallet?, token?
}

Status: 200 success · 400 missing query · narrative may fall back to mock if LLM unavailable

• Frontend → Next.js · TypeScript · Tailwind · Framer Motion
• AI → LLM API · agent / tool-calling ready
• Chain → Robinhood on-chain APIs / RPC (wired in later phases)
• Backend → Next.js API · PostgreSQL (planned)
• Deploy → Vercel

# .env.local
XAI_API_KEY=  # optional, SpaceXAI / xAI