Description
I need an AI Agent Developer to build an autonomous assistant that handles both phone calls and website chat, pulling answers from our documents and routing each query to whichever AI model — Claude or OpenAI — gives the best result. The system should work as one unified agent behind two channels: voice (via Twilio and Retell AI) and chat, both backed by the same RAG-powered knowledge base. When the agent isn't confident in an answer, it should automatically trigger a workflow (via n8n or Zapier) to escalate to a human instead of guessing. What I need built: An AI agent architecture (Python, FastAPI) powering both voice and chat channels from one backend Voice AI agent for real phone calls (Twilio + Retell AI) CRM integration with GoHighLevel for escalated conversations and lead capture Chat AI agent for website/messaging, sharing the same logic and knowledge base Multi-model routing between Claude API and OpenAI API based on query type RAG pipeline with a vector database (Pinecone) for document-based retrieval Automated escalation workflow (n8n or Zapier) for low-confidence responses Call transcripts and basic performance/analytics tracking Clean API layer and documentation for future integrations (CRM-ready) Ideal skills: AI Agent Development — building autonomous, multi-channel systems, not single-purpose bots Claude API and OpenAI API integration, with model-routing logic Voice AI (Twilio, Retell AI) RAG architecture and vector search GoHighLevel (GHL) integration Python,