The layer between customer feedback and your codebase. It collects from wherever customers talk — Slack, your site, transcripts, CSV exports — groups it into ranked priorities, and writes a codebase-aware spec for each one. When you ship, it notifies the customers whose feedback drove the feature, and their next message starts the next spec.
01What Annsa is
02Why it helps
Most teams already have the feedback. The problem is what happens to it: buried in threads, spread across spreadsheets, lost between support and product. Turning it into a decision takes days, and a decision into a spec takes hours. This compresses both into minutes.
- Feedback from 8 sources into one ranked list
- Priorities scored by volume, urgency, sentiment and revenue impact
- Specs generated with real file paths from your codebase
- Customers hear back when their request ships
03Set up your account
Onboarding asks your role, how you found Annsa, and your product in a sentence. That last one matters if you have no public site to scrape — it's what makes specs relevant. Change it later at Settings → Team & Account.
04Connect GitHub
Settings → Integrations → GitHub → Connect, then pick a repository. Annsa reads 9 signals for context and never writes code. Without it, specs still generate but the Files to Touch section is omitted.
05Bring feedback in
Pick at least one source; 10 pieces of feedback is roughly the point where priorities get meaningful.
Always-on — collect continuously once connected:
- Slack — Settings → Integrations → Slack, connect and select channels (
#customer-feedback,#support). Polled every 10 minutes; forwarded emails parsed automatically. - Surveys — Settings → Integrations → Surveys, pick a format (Float is quickest), paste the embed before
</body>. Arrives in real time.
Imports — one-time or recurring:
- CSV — upload a
.csvwith a feedback text column, map, preview, import. Good for NPS and support exports. - Google Sheets — connect Google Drive, select a spreadsheet, map the feedback column.
- Transcripts — paste or upload from Drive. Otter.ai, Fireflies, Grain and Whisper, up to 25 segments per call.
- Manual — + Add in the top nav, paste directly.
06See your priorities
Annsa classifies and groups automatically. Priorities rank by the User Growth lens by default — switch from the dropdown. Each card shows volume, type, trend, and a NEW badge if it appeared in the last 7 days. Click any priority to open its spec.
07Read a spec
| Section | What it contains |
|---|---|
| What to Build | One clear task statement |
| Why It Matters | Who's affected and the revenue context |
| Customer Voice | Real quotes from the feedback |
| Files to Touch | Suggested paths from your codebase (needs GitHub) |
| Done When | Exit criteria |
Click Copy to take the whole thing into Cursor, Claude Code or any editor.
08Or skip the copy-paste
Claude Code — run once:
claude mcp add --transport http annsa https://app.annsa.ai/mcp
First run opens a browser for OAuth. Then ask: "what are my top priorities?" or "get the spec for [priority name]."
09Ship and close the loop
Open the spec and click Mark as Shipped. The Share Back dialog opens — review the recipients, choose email, survey banner or both, and send. The spec moves to Shared, and new feedback on that theme feeds a V2 spec automatically.
10What's next
Invite your team at Settings → Team, add more channels (Slack, Surveys, transcripts, API), and connect MCP so you can work from Cursor or Claude Code. Questions go to support@annsa.ai.