Description
We're planning a connected internal AI layer over time. This is the first paid pilot module — deliberately small and self-contained so we can see the approach work before we expand. If it lands, there's ongoing work on the next modules. Right now our team manually scans our CRM and Google Sheets every morning to work out what needs attention. We want one AI agent that does that first pass for us. What you'll build •One AI agent (Claude / Anthropic SDK) that pulls from our CRM + Google Sheets via API/MCP •A short RAG step over ~[X] of our internal docs so answers are grounded in our real data •Produces a clear, prioritized daily action list — what matters, surfaced, not raw reports •Wired in n8n, delivered to us in Slack each morning •A short handover doc so we can adjust it ourselves You should have Python (FastAPI a plus), Claude/OpenAI API experience, RAG, API/MCP integration, n8n or Make, prompt engineering — and a live example of an AI agent or automation you've shipped. Comfort with Claude Code and MCP is a strong plus. Out of scope for this pilot No multi-business rollout, no autonomous actions without a review step — just this one reliable daily agent.