Zendesk CSAT Follow-up Agent with Claude Code

One low rating is not one read. It is the whole exchange in order, the exact reply where the customer's tone changed, and then a second sweep across every other ticket that person has ever filed. Claude Code is reliable on long structured tasks, and this one is long before it is difficult: several reads, a judgement about a single moment, then eight lines written for a busy colleague.
Loading preview…
Free to start · guided setup · switch agents any time.

Live demo

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

Open

Yuki 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.

What you'll need

Set up in minutes

Using this template drops you into a guided setup. It asks exactly this, nothing else:

  1. Connect Zendesk

    One sign-in. The agent acts through your account, scoped to what this template uses.

  2. Connect Slack

    One sign-in. The agent acts through your account, scoped to what this template uses.

  3. 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.

  4. Runs on Claude Code

    Preselected for this page — connect your Claude Code account during setup, or switch to NoClick's built-in models with one click.

  5. Watch it handle a test run

    A staged conversation against a simulated world — then it’s live.

A considered pairing

Why Claude Code for this agent

The turning point is one line

Finding the message where it went wrong means keeping the whole thread in view while judging one reply against the ones around it. That reading gets worse the faster it is done.

Third time, not first time

Pulling the requester's other tickets and naming the earlier numbers turns an incident into a pattern. It is an extra hop the run has to take after it already believes it has an answer.

Good to know

Before you fork

Does reading every ticket in full make it slow on a bad week?

It is thorough by design, so a run takes longer than a summariser would. What it faces is your low ratings rather than your ticket count, and scores your rules count as good stop at the check within seconds. Thirty bad ratings in a week is thirty careful reads, still less time than one person spends on five of them.

Other editions

Run it with a different agent

OpenAICodexopencodeopenclawhermesMore Claude Code agents

Make it yours

Put Zendesk CSAT Follow-up Agent to work on Claude Code.

Free to start. Guided setup, a staged test run, and a workflow you can change at every step.