Getting started
Engram CaaS documentation
Everything technical about Engram in one place: how to load documents, how to ask them questions from your own tools, and how to run the workspace around them. Pick an area on the left, or start with the guides below.
What Engram is
Engram turns a set of documents into a document base: the documents are read once, compiled into resident memory on our serving hardware, and from then on any question is answered straight from that memory, with the documents it drew on named as sources. You do not run a retrieval pipeline, chunk anything, or pay per document to keep it ready. You load documents, wait for onboarding to finish, and ask.
Three surfaces sit on top of every document base, and each has its own section here:
- MCP: expose every document base in your workspace as a tool to Claude, Cursor or any MCP-capable client, so your agents query them instead of ingesting the documents themselves.
- CLI:
engrampushes folders, manages document bases and keys, prints the MCP config block, and scripts cleanly with--json. - REST API: the same operations over HTTPS with an API key, for anything you build yourself.
Start here
Which path to use
| You want to | Use |
|---|---|
| Load a few files from your laptop | The app's onboarding wizard |
| Load a large folder, or keep a folder in sync | CLI push |
| Keep a Drive folder or a SharePoint library in sync | The Sources section of the Documents tab |
| Keep an S3 bucket in sync | The CLI |
| Let an AI assistant or agent answer from the documents | MCP |
| Build your own integration | REST API and webhooks |
How to read these pages
Code blocks have a copy button. Placeholders look like <your key>. Every example targets the production API at https://api.engramdynamics.org; the app is at https://app.engramdynamics.org. Where a feature belongs to a paid plan, the page says which one and links to plans and allowances, which is the one place the numbers live.