Skip to content

Start here

Getting started

Run Sergo on one machine with Docker Compose.

Sergo runs as a set of containers: Postgres, the API, processors that read files, and the web console. One command starts them all.

  • Docker with Compose
  • A few GB of memory for Docker. LibreOffice, which converts office files, uses most of it.

Sergo is in pre-release, and we send the source on request: use Deploy on-premises on the home page. It’s free for personal use, for non-profit organizations (other than government and public bodies), and for evaluating it at your company for up to 32 days. Any other use, such as running it in production at a company, needs a commercial license. LICENSE.md in the source has the full terms.

From the folder you unpacked the source into:

Terminal window
docker compose up -d --build

When the containers are healthy, open the console at http://localhost:3000 and sign in as admin. The first password comes from SERGO_BOOTSTRAP_ADMIN_PASSWORD in .env. If you left it out, Sergo made one up and printed it once in the api log:

Terminal window
docker compose logs api | grep "first admin"

Either way, the console asks you to choose your own password before anything else. If you forget it later, set a new one from the server:

Terminal window
docker compose exec api python -m sergo.users set-password admin
  • Postgres in a container, holding everything except the uploaded files, which live in a Docker volume.
  • Keyword search on Postgres. Admin › Search index shows one definition, default, that mixes two signals at 0.5 each: Keywords (words that appear in the chunk) and Rare words (the same keywords, weighed by how rare they are across all chunks).
  • No embedding model, so no search by meaning yet.
  • Basic processing works right away. Standard (AI) needs a model provider, which an admin adds in Admin › Models. See Processing.

Everything listens on 127.0.0.1 only. To reach the console from other machines, put a reverse proxy in front of port 3000.

  • Upload a few files and process them. See Processing.
  • To let an AI app search your documents, create an agent key. See Agent keys and MCP.
  • To add search by meaning, use your own Postgres, add OpenSearch, or run more processors, see Configuration.