Architecture
XanLens is built for speed and reliability. Stateless API routes handle audit orchestration, while Redis provides job state and caching.
Stack
- Next.js 16 on Vercel — frontend and API routes
- Upstash Redis — job state, audit data, caching
- Gemini API — primary AI engine + LLM judge + content generation
- Grok API (xAI) — knowledge engine
- DeepSeek API — knowledge engine
- Gemini with Google Search — grounded discoverability analysis
- Tavily API — AI-native search with real-time web data
- x402 protocol — HTTP-native USDC payments on Base
- Coinbase Smart Wallet — gasless payments via CDP Paymaster
Request Flow
User/Agent → Auto-Detect (brand, industry, competitors)
→ Submit Audit → Payment (USDC / coupon / x402)
→ Execute → [Gemini, Grok, DeepSeek, Gemini Grounded] (parallel)
→ LLM Judge validates each response
→ Score + Findings → Content Fixes Generation
→ Dashboard (approve / reject / edit fixes)
Engines
Currently live: Gemini (knowledge), Grok (knowledge), DeepSeek (knowledge), and Gemini Grounded (discoverability via Google Search). ChatGPT, Claude, Perplexity, and Meta AI are supported but currently paused — coming online as API capacity scales.
Design Principles
- Fast — parallel engine queries, edge-deployed
- Portable — works from any HTTP client, curl to Claude
- Transparent — scores show exactly what each engine said
- Agent-first — REST API, MCP server, OpenClaw skill