Two refinements from the first Starko AI runs in the wild. Comments the agent posts on a ticket — the "Handed to Starko AI" note and the outcome summary — now carry the automated workflow icon and badge instead of showing up under the avatar of whoever triggered the run, so it's always obvious which updates came from a person and which came from the system.
And when a run fails to launch, the ticket now tells you why. Instead of a generic "agent stopped without reporting a result", you get the actual launch error posted as a comment (with any sensitive details scrubbed), so you know whether to retry or reach out to us.