An engagement is a meeting. More precisely, it’s the structured form of one specific conversation: title, date, summary, sections, decisions, actions, attendees. Every engagement belongs to exactly one journey.Documentation Index
Fetch the complete documentation index at: https://docs.runwita.com/llms.txt
Use this file to discover all available pages before exploring further.
The shape of an engagement
When the AI extracts a transcript, you get back this structure:| Field | What it is |
|---|---|
| Title | A descriptive title written by the AI from the actual content, not the calendar name. “Joule SCI Daily Check-In: GA timeline + LoB asks”, not “Daily standup”. |
| Date | The date the meeting happened. Used for journey timeline ordering and as the anchor for resolving relative due dates in actions. |
| Type | meeting, email, or (uncommon) other source kinds. Determines which extraction prompt the AI used. |
| Summary | Bullet-point recap of the highlights. Read-it-in-twenty-seconds version. |
| Sections | The detailed notes, organised by topic-of-discussion. Each section has a heading and a body paragraph. A 30-minute meeting typically yields 3 to 5 sections; an hour yields 5 to 8. |
| Decisions | Things that were explicitly agreed. Each decision has the text and an owner (the person responsible for what was decided). |
| Actions | Concrete follow-ups with an owner, an optional due date, and a confidence flag for the date (explicit / inferred / none). |
| Attendees | People who were in the room (or thread). Pulled from speaker labels in transcripts and From/To/CC in emails. |
What an engagement is NOT
It’s tempting to think of an engagement as the whole meeting, the raw transcript and all. It isn’t. The transcript stays attached to the original inbox item. The engagement is the distilled output: structured, searchable, summarised. You can always click back to the source if you need the raw text, but the engagement is what shows up in journey timelines, topic histories, and search.How engagements relate to topics
This is where things get interesting. The sections in an engagement aren’t standalone, they get matched against the journey’s existing topics when you save. Each section becomes a touchpoint on either an existing topic (if the AI decides this is the same thread continuing) or a brand-new topic (if it’s genuinely new). Worked example. Imagine a journey with these existing topics: “Licensing model”, “Migration timeline”, “Stakeholder alignment”. You save a new engagement with these sections:- Singleton Joule licensing options → matches existing “Licensing model”, becomes a touchpoint, status flips to whatever the AI thought happened (decided, blocked, etc).
- Q2 timeline review → matches existing “Migration timeline”, new touchpoint.
- New OEM partnership → no existing topic matches, creates a new topic “OEM partnership” and seeds it with this touchpoint.
Merged engagements
Sometimes one logical meeting produces several inbox items, you joined from two devices and got two transcripts, or there’s the recording plus your handwritten notes plus a screenshot of the whiteboard. You can select multiple inbox items and run merged extraction: the AI gets all of them at once and produces a single engagement that draws from every source. The first transcript becomes the spine (its date anchors the engagement). Other transcripts are stitched in chronologically. Non-transcript items (notes, OCR, emails) ride along as supporting evidence, the AI uses them to enrich the summary and topics but doesn’t treat them as dialogue. Merged engagements get a higher AI output budget (32K tokens vs 16K for single) because the input is intrinsically bigger. See AI tiers.Editing after save
Every field on an engagement is editable after you’ve saved it:- Click into the engagement on the journey page.
- Title, date, summary, sections, decisions, actions, attendees, all editable inline.
- Changes save immediately, no save button.
Deleting an engagement
Engagements can be deleted from their detail page. Deleting an engagement also removes its touchpoints (so the topic states on the journey re-compute), its decisions, actions, and attachments. The original inbox item is unmarked as processed, which means it’ll show up as a fresh capture again. That’s intentional, so you can re-extract.What’s next
Topics
The threads that span engagements.
AI tiers
Frontier vs Workhorse, and when each runs.