Anchor AI Support Agent
RAG-powered AI customer support agent with citations and human handoff.
Is this your project? Ask us to take it down
Gallery
About
An AI customer-support platform that answers questions only from your own knowledge base, with numbered citations. If it isn’t confident enough, it declines instead of guessing.
Hybrid retrieval (Pinecone vector search plus BM25 keyword search merged with Reciprocal Rank Fusion), a reranking stage, multi-turn memory with query rewriting, intent classification across 7 support categories, sentiment tracking, a knowledge-base crawler with change detection, a human-escalation queue with signed webhooks, quality scoring, cost tracking and PII redaction. The Next.js dashboard has chat, knowledge-base management, a retrieval inspector, session analytics and escalation views. Built with FastAPI, Next.js 15, Pinecone and OpenAI. 31 automated tests.
Comments (3)
how does the hybrid retrieval work with the citations? what kind of architecture is behind it?
wait, so it actually refuses to answer if confidence is low? doesn’t that just burden the human handoff queue way more?
citations? interesting choice. does it really improve trust?
Related Products
Typist
Transcribe audio to text in seconds.
rapym
TKCORE AI
All-in-one AI platform featuring TkCore-V5.5-Pro & top LLMs
Textsight.ai
TextSight.ai detects AI-like writing, gives a 0–100 Authenticity / Humanization
ai image prompts
Copy-paste AI image prompts that actually work.
ChatFlow
AI chatbot for your website — live in minutes
ComingUp