Your community tells you exactly what's broken and what they want — in a firehose no one can read. SummaryBot's Intelligence Reports rank the feedback, count it from the real messages, and hand you a prioritized report you can act on.
/summarize/ask/exportmessagesIntelligence Reports rank every theme by weight and break out praise versus complaints per feature, so you know what to fix first. You walk into a roadmap review with the top issues and requests already sorted, instead of a channel you've been meaning to read. It's the difference between having feedback and understanding it.
This is the part your team can actually stand behind. When a report says "14 people reported this bug", that's a real tally computed deterministically from the message data — not a figure the AI invented. The model classifies and explains, but the counts, rankings, and reaction weights come from code that reads the messages, so the numbers hold up in a standup or a board meeting.
Recurring digests tag every theme New, Recurring, or Handled, so you see what changed instead of re-reading the same issues. Reaction-weighted analysis surfaces what the community actually cares about rather than just what's loudest, and forum groups consolidate a busy bug-report forum into one deduplicated section of the report.
Ask for "all feedback since v2.3" and get a report scoped to exactly the window that matters for your release. Game studios can see how players really reacted to a patch; product teams can measure a launch. Reports land in Discord and your inbox, and HMAC-signed webhooks pipe the intelligence straight into your own tools.
Counts are computed deterministically from the actual messages, never estimated by the AI. The model classifies and explains each theme, but every "X people reported this" tally, ranking, and reaction weight comes from code that reads the real data — so the numbers hold up in a roadmap review or board meeting.
Yes. Release-anchored reports let you ask for "all feedback since v2.3" and scope the analysis to exactly that window. Game studios use it to see how a patch landed; product teams use it to measure a launch against the release calendar.
No — that's what digest memory prevents. Each theme is tagged New, Recurring, or Handled, so a weekly report shows what changed since last time instead of resurfacing the same backlog. Reaction weighting keeps the signal that matters most at the top.
Reports arrive in Discord and by email, and can be pushed through HMAC-signed webhooks into your own dashboards and tools. You can also pull the raw messages with full per-emoji reaction data using /exportmessages for your own analysis.