Zendesk CSAT Follow-up Agent with OpenCode
The way this goes wrong is not a missed rating. It is a confident note naming a cause the thread never showed, repeated in a review a week later as though somebody had checked. With OpenCode the model making that judgement is your pick, which matters because restraint under an ambiguous exchange varies far more between models than accuracy on the obvious ones.
Watch it work before it's live
Run a staged conversation — no account needed. The agent handles it for real while a simulated world answers its tool calls; nothing touches real accounts, and nothing is actually sent.
CSAT 1/5 on ticket #48213
OpenYuki Tanabe <#48213>
Three different people asked me to send the same screenshot, and the third one asked me to explain the whole problem again from the beginning. Eleven days to change one setting. I have no confidence it would go any better next time.
Set up in minutes
Using this template drops you into a guided setup. It asks exactly this, nothing else:
Connect Zendesk
One sign-in. The agent acts through your account, scoped to what this template uses.
Connect Slack
One sign-in. The agent acts through your account, scoped to what this template uses.
Followup rules
Which scores you want worked and what should happen to them: the threshold that counts as low, who gets the Slack line, the accounts or plans that always matter, and what a reasonable recovery step looks like at your company.
Runs on OpenCode
Preselected for this page — connect your OpenCode account during setup, or switch to NoClick's built-in models with one click.
Watch it handle a test run
A staged conversation against a simulated world — then it’s live.
Why OpenCode for this agent
Say the cause is not visible
Where the thread does not explain the score, the note has to say so and name the one question that would answer it. Weaker models fill that space with something plausible.
The judging happens in one place
Which scores get worked, who hears about them, and what counts as a recovery step all live in your follow-up rules rather than in the model. Changing engines moves none of it.
Before you fork
How do I see whether a model reads a rating fairly?
Run the staged Two Stars, No Comment rehearsal, which is the hard case: a low score with no written comment and a fourteen hour overnight gap sitting in the ticket. A model that names the gap is reading properly. A model that invents a rude reply is the one to leave behind.
Run it with a different agent
Put Zendesk CSAT Follow-up Agent to work on OpenCode
Free to start. Guided setup, a test run against staged conversations, and it's live.