Journey matching runs after extraction. The AI looks at the meeting content and picks one of your existing journeys, with a confidence label (high, medium, low, none). When it gets it wrong, you’ll see one of three failure shapes.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.
”It matched the wrong journey”
The AI picked an existing journey, but it’s not the right one. Usually because two journeys have similar customer names and the AI weighted the customer signal too heavily. Fix: on the Review screen, click Change in the Journey panel. Pick the right journey from the list, or click + New to create one inline. Then save normally. This is a one-time correction, the saved engagement is on the right journey. The AI won’t suddenly learn from this and get smarter (there’s no per-user feedback loop yet), but the immediate fix is fine. If two specific journeys keep getting confused with each other, the long-term fix is to make their use case fields more distinctive. “Acme Corp, Renewal” and “Acme Corp, Q4 expansion” are easier to disambiguate than “Acme Corp, Working sessions” and “Acme Corp, Calls”.”It said ‘no match’ when there’s clearly a match”
The AI returnedconfidence: none and you can see the right journey is in your list. Usually because the journey’s use case is too vague for the matcher to connect to the meeting content.
Fix on the Review screen: just pick the right journey manually (Change → select). Same as above.
Fix going forward: rename the journey to be more specific. The matcher works on word/concept overlap between the meeting and the journey’s identity (customer + use case + summary). If your use case is “Engagement” or “Working with them”, there’s nothing for the matcher to connect to. “Migration to BTP and post-launch governance” gives it something to work with.
”It picked a journey when this should be a new one”
The AI matched to an existing journey because the customer name is shared, but this meeting is genuinely about a new use case that deserves its own journey. Fix: on the Review screen, click + New journey. Fill in the customer (often pre-filled correctly) and a fresh use case. Save. The new journey is created and the engagement lands on it. You can also tell the difference between “close enough, file it” and “this needs its own thread” by reading the AI’s reasoning, hover the journey-match chip on the inbox row, and the tooltip shows the AI’s brief explanation.How to tell journey match is going wrong systematically
Two patterns suggest the matcher needs a model upgrade:- Frequent low-confidence matches when you can clearly see the right journey. The AI is saying “I don’t know” when the answer is right there, smaller models do this when they fail to see the customer/use-case overlap.
- Confident matches that are wrong. The AI is saying “definitely this journey” with high confidence, and it’s wrong, smaller models can over-commit.
Editing journey assignment after save
Once an engagement is saved to a journey, you can’t currently move it without deleting and re-saving. Editing journey assignment in place is on the roadmap. Workaround for now:Delete the engagement
Open the engagement → delete. The original inbox item is automatically un-marked as processed and shows up in the Inbox again.
Reprocess
From the Inbox, click into the now-back item, click Process with AI, set the date if needed, run extraction.
What’s next
Concepts: journeys
Why specific use case names make matching better.
Settings: models
Upgrade your Workhorse tier if matching is consistently off.