Live demo running on Workers AI

Ask the NeuronShield brain.

A RAG system over everything we've published — case studies, blog, playbooks. Built on Cloudflare Workers AI + Vectorize. The same architecture we ship for production RAG clients.

Try:
How it works

The stack behind this demo

  1. 1. Embed. Question → 384-dim vector via @cf/baai/bge-small-en-v1.5 on Workers AI.
  2. 2. Search. top-k cosine similarity against a Vectorize index of every case study, blog post, and playbook on this site.
  3. 3. Generate. Retrieved chunks + the question → @cf/meta/llama-3.1-8b-instruct with a strict "answer from context only" prompt.
  4. 4. Cite. Sources surfaced alongside the answer for transparency.
Same architecture pattern we ship for production RAG clients — see the full case study →
Want a system like this in your stack? Email us →