Agent Engine
- Dror Margalit
- Nov 12
- 1 min read
What if you could have an AI agent that can talk on your behalf? It understands your knowledge, your tone, and your behavior.
Agent Engine is an experiment that turns people's expertise into a conversational AI. They can add content, choose a voice, publish, and let people ask questions anytime, on a site, blog, or app.
Try it
How it works
Agent Engine chunks and embeds files into a vector DB. Then using Retrival Augmented Generatioin (RAG), it brings back the most relevant passages into a Large Language Model (LLM) that composes an answer and cites the sources it used. If the evidence is thin, it tells the user.
The end user doesn't need to know how to prompt the AI. They simply need to chat with it and engage with it based on the instructions and content provided by the agent facilitator.

Your knowledge
Agent facilitators can drag in PDFs, decks, transcripts, CSVs other files. Agent Engine chunks, vectorizes, and keeps everything organized for grounded answers.

Your style
Facilitators give the agent a name, tone, and guardrails. Instructions stay editable so you can iterate fast.

Anywhere
Publish with a link or embed the chat on sites, LMS pages, or portals. It stays synced with your latest changes.
Create your own agents: https://agent-engine.com/




Comments