Intercom Support Agent with Codex
Live chat is judged on how quickly the first useful sentence arrives, and most of what arrives is a question your playbook already answers in three lines. Codex is fast on concrete, well-scoped requests and needs no runway to produce one to three sentences. Anything outside the playbook becomes a note and a handover instead of a reply.
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.
Adding a teammate to the workspace
OpenSiobhan O'Rourke <siobhan@rathmoredesign.ie>
Hey, how do I get my colleague into our workspace? I only see my own name under members and we are on the Studio plan.
Set up in minutes
Using this template drops you into a guided setup. It asks exactly this, nothing else:
Connect Intercom
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.
Support playbook
The answers you are happy to give without a human checking - your common questions, how-to steps, plan limits, and known issues, in your own words. The agent replies only when the answer is in here.
Escalation rules
The conversations a person must own - refunds, billing, account access, security, cancellations, angry customers, named accounts, whatever you decide. The agent acknowledges these and hands them over instead of answering.
Runs on Codex
Preselected for this page — connect your Codex 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 Codex for this agent
Short answers, fast
The instruction is one to three sentences with no greeting boilerplate. A harness suited to concrete tasks does not need talking down from writing an essay.
Looks up the person first
The contact record and any earlier conversation are fetched before it types, so a returning customer is not asked to explain themselves twice.
Before you fork
Can it reply to something it should not?
It may reply only to what the playbook covers, so the playbook is your control surface, and money, access and cancellation escalate whatever the playbook says. The realistic failure is a question sitting at the edge of a playbook section that gets answered where a person would have escalated. Run your borderline cases through Test Run and tighten the wording wherever it overreaches.
Run it with a different agent
Put Intercom Support Agent to work on Codex
Free to start. Guided setup, a test run against staged conversations, and it's live.