Tell us about the last time your agent failed because it lacked context.
We are collecting short failure postmortems from builders and operators running real AI agents. One concrete story is more useful than ten abstract opinions.
We care about one real workflow, not your whole architecture.
The useful detail is where the missing context actually lived: schema, docs, Slack, runbooks, tickets, or someone's head.
We also want the workaround and the cost: time, trust, support load, or business risk.
What qualifies
Live or near-live agent work
Internal copilots, support agents, workflow agents, data agents, coding agents, or any autonomous system running against real company context.
One concrete failure
Example: wrong source of truth, missing business definition, policy exception, stale documentation, or missing workflow state.
Bounded follow-up
If your story is especially relevant, we may ask for a 20-minute follow-up. This is research, not a sales process.