This page explains the transcript pipeline and the two summary modes OmegaBot supports.
The key distinction is simple:
- transcripts are the stored record of source messages
- summaries are derived output generated from those transcripts
Messages move through the system in this order:
- Messages Raw messages are collected from the source, such as Discord.
- Transcript Messages are normalized, ordered, and grouped with timestamps and metadata.
- Summary The transcript is summarized into a shorter human-readable form using the selected summary mode.
This means the transcript is the durable source record, while the summary is a generated view of that record.
- OmegaBot uses IANA timezone identifiers such as
America/New_YorkandUTC - timestamps are normalized during transcript generation
- display timezone can change without regenerating the underlying transcript
- heuristic or rule-based summaries
- fast and deterministic
- no external API calls
- useful for quick overviews or offline-friendly behavior
- uses a language model to generate summaries
- produces more contextual and natural phrasing
- requires external API access and the related config
- useful when you want richer narrative summaries
- transcripts remain the source of truth
- summaries can be regenerated without reprocessing the original message source
- changing summary mode affects summary style and quality, not transcript integrity