Petals MemoryInstall

Summary

Ingests a transcript into Petals memory, preserving speaker labels and optional relative timestamps when you provide pre-segmented utterances. Choose raw mode to send full text for server-side segmentation, or segmented mode to send utterances directly; set the transcript ID, coverage timestamp, and scope.

Inputs

• Transcript ID (required) — Stable transcript ID in your namespace.
• Occurred At (required) — ISO 8601 timestamp for when the transcript content occurred; the node converts valid dates to UTC ISO format.
• Mode — Choose Raw (default) to send full transcript text for server-side segmentation, or Segmented to provide utterances.
• Text (required) — In Raw mode, the full transcript text.
• Utterances — In Segmented mode, add utterances with a speaker label, content, and optional relative timestamp.
• Scope — Personal by default. Choose Reference for curated material surfaced separately from personal queries.

Output shape

One or more n8n items containing the API response as JSON.

The node sends one POST request per input item to Petals’ transcript-ingestion endpoint. If the response is an array, each element becomes a separate output item; otherwise the response becomes one item. With Continue On Fail enabled, errors are returned as items containing an error message.

Examples

Example 1: Store a transcript as raw text in personal memory

Set Transcript ID and Occurred At, keep Mode as Raw, enter the full transcript in Text, and use Personal scope.

Example 2: Store speaker-attributed utterances as reference material

Set Transcript ID and Occurred At, choose Segmented mode, add utterances with speaker labels and content (and optional relative timestamps), and select Reference scope.

Links

• n8n community nodes installation guide