Atlassian previews 'Classify with Rovo', a Jira and Confluence automation step built on Jev
publicado (OK do Fabricio 19/10 BRT).
Atlassian’s Studio Automations team has opened an early access program for Classify with Rovo, a new automation step powered by TypeSafe’s Jev. Ben Costello announced it in the Atlassian Community on 15 Oct (UTC) and says the team showed it on stage at Team ’26 Europe.
The step offers three kinds of classification, mapped to Jev’s question types:
| Atlassian name | What it does | Jev type |
|---|---|---|
| Check | A true/false question about the text | noul |
| Categorize | One of two or more categories | choice |
| Score | One of two or more ranks or ratings | score (Atlassian does not name it) |
One step can run several classifications on the same text. The examples are routine Jira and Confluence work: priority, request or issue type, which team gets a work item, which label a page needs, and whether a page should be archived or deleted.
In the comment thread, Costello adds a few details:
- Each classification returns all options with their probabilities and an overall confidence, plus a single answer (the most likely option), as structured smart values that later automation steps can use.
- Pricing has not been announced.
- Atlassian is “evaluating others” among the decision models released since Jev.
Why it matters. This is the first Atlassian product step built on a decision model, and the probabilities reach the user instead of being hidden behind a label. Commenters asked for a way to send uncertain cases to a person and to test against existing work items before high-impact actions such as deleting content. The confidence output makes the first possible; the second is not described yet. It is early access, so we have no usage, latency or cost figures to report. Added to the distribution list on the Jev page; no score changes.