Meeting Notes & Summaries OpenAI-Compatible · LAN / Localhost

Generate meeting notes with a local LLM

Connect Stellar Recorder to an OpenAI-compatible chat model server running on your PC or private network, enable notes deliberately, and verify every generated point via clickable transcript citations.

Opt-In Local LLM Summary Architecture
Meeting Notes & Summaries OpenAI-Compatible · LAN / Localhost
01 · Model Host

Run Local Chat Server

Start your OpenAI-compatible server on localhost or an internal workstation you inspect.

02 · Configure

Enter URL & Exact Model

Save endpoint address, port and exact model ID inside Settings → AI summaries.

03 · Enable

Turn On Notes Explicitly

Saving endpoint credentials keeps notes off until you explicitly activate note generation.

04 · Verify

Inspect Cited Passages

Click linked timestamps beside decisions and action items to audit output against audio.

Stellar Recorder - Cited Decisions, Action Items & Formatted Copy
OpenAI-Compatible LLM · Citations
Action items with expandable transcript quotes and Copy formatted action in Stellar Recorder

Choose an OpenAI-compatible model server

Fresh Stellar Recorder profiles keep AI notes disabled. Any local runtime or LAN gateway exposing a standard OpenAI-compatible chat completions endpoint works - check your model server documentation to match VRAM/RAM limits. Remember that even a LAN IP only describes transport; verify the host’s own logs and retention.

Check every note against source lines

Smaller local quantized models trade reasoning capacity for speed. Because every summary bullet, action item and decision in Stellar Recorder attaches a source timestamp link, always verify names, prices, dates and commitments against the transcript passage.

Configure local LLM notes in Stellar Recorder

  1. Open Settings → AI summaries inside Stellar Recorder.
  2. Enter the full endpoint URL (including port) of your running OpenAI-compatible server.
  3. Enter the exact model ID string exposed by that server and save settings.
  4. Toggle automatic AI notes on explicitly (saving credentials alone does not enable automatic notes).
  5. Open a finished recording, generate or inspect notes, and test click-to-seek links.

Summary connection troubleshooting

  • Notes stayed empty after saving settings: verify automatic notes are switched on and click generate on finished recordings.
  • Model not found errors: match the exact model string returned by your inference server `/v1/models` list.
  • Connection refused: confirm your local LLM server process is active and bound to the configured port.