We review 20+ customer calls weekly. Sembly summaries are good, but actionable items were getting lost in Slack. Needed them in Jira, where the engineering team lives.
Built a simple pipeline:
* Sembly webhook triggers on "Meeting Ended" → sends JSON to a lightweight Flask app.
* App parses the "Action Items" section from the summary.
* For each item, we extract:
* Owner (from assignee tag)
* Deadline (if present in text)
* Raw item text
* These are mapped to Jira ticket fields via the REST API.
* Key metrics after 3 months:
* 85% reduction in manual ticket creation for call follow-ups.
* Avg. time from meeting end to ticket creation: 2.1 minutes.
Biggest win? It forced clarity in action item wording during calls. If it's not clear enough for the parser, it's not clear enough for the dev.
Prove it with a benchmark.
That's clever. I've been looking at automating follow-ups from our standups. Do you use a specific Jira project for all these auto-created tickets, or do they get routed based on the meeting title or content? I'm worried about creating ticket sprawl.
Excellent point about forcing clarity through the parser's constraints. That's a benefit we often overlook when designing automation - it doesn't just streamline a process, it enforces a discipline.
In my procurement work, I see similar patterns with contract lifecycle automation. When a system requires a clear "parties" field and a defined "termination date" before routing for signature, it forces legal teams to be specific in early drafts. The mechanical requirement eliminates vagueness.
Have you found that this clarity in action items has changed the dynamic of the calls themselves? I'm curious if participants now self-correct their wording in real-time, knowing it will become a ticket.
null