Skip to content

What are AI-Glance Summaries?

AI-Glance Summaries are short, AI-generated 3–5 word summaries that appear on every mention card in your inbox. They give you instant context without opening the issue or page.

Each mention displays a one-line AI summary describing what the mention is about. For example:

  • “Review pull request changes”
  • “Database performance issue”
  • “Approve Q3 budget proposal”
  • “Deploy to production today”

These summaries are generated by the same AI that classifies Action, Urgency, and Impact — they’re based on the full context of the comment and the issue or page it’s in.

AI-Glance Summaries appear:

  • In your Krina inbox on every mention card
  • In the Krina panel in Jira issues and Confluence pages
  • In the Follow-ups view alongside your follow-up date

They help you scan through mentions quickly and prioritize what to tackle first, without reading the full text.

  • Quick triage — Scan multiple mentions at a glance without opening each one
  • Accurate context — The summary reflects the actual ask, not just a keyword match
  • Searchable — The summary text is indexed alongside the full comment, so you can search for concepts even if those exact words didn’t appear in the original text

AI-Glance Summaries are generated at the same time as the classification (Action, Urgency, Impact). They’re as accurate as Krina’s classification — if a mention is miscategorized, the summary may also be slightly off. If you notice inaccurate summaries, they usually resolve when you re-classify the mention or when Krina’s AI model is updated.

The summary is NOT the beginning of the comment. It’s an AI-generated extract that captures the essence of the ask. The full comment text is still stored and fully searchable — the summary is just the headline.

AI-Glance Summaries have a few edge cases where they may be less precise:

  • Very short comments (one or two words, emojis only) — The summary may be generic or unhelpful
  • Ambiguous comments — Comments where the ask is implied rather than explicit may produce inaccurate summaries
  • Complex requests — Multi-part requests may only capture part of the ask
  • Sarcasm or irony — The AI may interpret sarcasm literally

In these cases, the full comment text is always available to read. If you find a summary consistently inaccurate, re-classifying the mention (edit and save the comment again) may improve future summaries.

Privacy PolicySivect