One call, start to finish.
One discovery call, from the prep that reads your backlog to the tickets sitting in Linear before you've left the meeting.
Read the invite: Northwind - Bulk import discovery
Searched your memory - 2 earlier conversations
Linear - 4 items
Notion - 2 items
Slack - 6 items
It reads your backlog, not just your calendar
Caspi pulls the open Linear and Jira issues, the relevant Notion pages and what this customer said last time, then tells you what's still unresolved before you join.
Duration is identical across all failed jobs regardless of row count, which indicates a fixed request timeout rather than a volume ceiling.
Every failure is a timeout at exactly 30s - not a row-count limit.
It's the timeout, not the volume. Reframe the ask around retries.
It reads the log they just shared
The customer shares an error export. Caspi finds the pattern in the background: every failure stops at exactly 30 seconds regardless of row count - a timeout, not the volume limit everyone assumed.
Do we already have a ticket for this?
Yes - GAR-142, open 5 weeks, linked to two other accounts. Mapping is GAR-118, closed.
“Do we already have a ticket for this?”
Ask without leaving the call and Caspi checks Linear, Jira and GitHub - so you attach evidence to the issue that exists instead of opening the same one twice.
Drafted and ready to send
Subject
Northwind bulk import - what we found
Tickets and specs, not a summary
Caspi drafts agent-ready specs for Linear, Jira or GitHub and the decision log for Notion - each checked against what already exists before anything is created.
LINEAR
GAR-142: add evidence - 30s timeout, 40k rows, 2×/week
NOTION
Discovery: Northwind bulk import - decision log
It extends the ticket instead of duplicating it
The evidence attaches to the Linear issue that already existed, and the decision log appears in Notion. You approve each one, and anything already covered is shown to you rather than created again.
