← All AI work
Support insights in Slack, every day
AI reads every support chat and posts what business owners struggle with to Slack at the end of each day.
- My role
- Set up the AI agents
- Built with
- Intercom · Amplitude · MCP · Slack
- Read by
- Stakeholders, in a Slack group
01 · Research
A week between a customer problem and the product team
Customer problems reached product in a weekly sync call with the support lead. By then, an issue could be a week old.
02 · Ideation
Pair what customers say with what they do
A chat tells us the problem; usage data tells us where it happened. Reading both together gives a 360 view of each business owner: their business type, and the steps where they got stuck.
03 · Solution
Buckets for every issue, and the usage behind it
- Sort
- Each chat goes into a bucket, using the rules we already had for known cases
- New issue
- A case that fits no bucket starts a new one
- Summary
- Posted to stakeholders in Slack at the end of each day, with Amplitude events behind each issue
04 · Success
Product decisions on live customer signal
Weekly → dailyHow often support issues reach the product team
Same-day signalProduct and design decisions run on what customers hit this week, not last week