Use case · Support
The queue, handled before standup.
Cedric reads every ticket, drafts every reply, and asks before anything goes out. Your team keeps the judgment calls.
$50 in free credits when you sign up. No card needed.
- Support work is mostly triage and typing, and both happen at hours nobody wants to work.
- Cedric reads every ticket as it lands, classifies it, routes the urgent ones with context attached, and drafts the rest.
- Every reply waits for a human yes. Cedric proposes, your team approves, and the customer never sees anything you did not sign off.
- It works from the tools you already have: your helpdesk, your billing system, your docs, with scoped and revocable access.
- The weekly support digest stops being someone's Friday afternoon and becomes a scheduled job.
The queue is never the hard part
The hard questions are the part support teams like. The cost is everything around them: reading forty tickets to find the three that matter, digging up the order, pasting the same explanation, checking the queue at 7am.
A front-of-queue chatbot does not touch that half; it only deflects what your docs already answer. Cedric works the half that lands on your team.
Triage
Overnight tickets, read and sorted by morning.
Maya Chen9:14 AM
@Cedric anything urgent in the overnight queue?
CedricAPP9:15 AM
Thirteen came in since 18:00. Two look urgent: a payment failure that hit checkout for four customers, and an enterprise admin locked out of their workspace. I posted the payment one in #engineering with the Stripe error and the affected accounts. The lockout has a reply drafted with the workaround. The other eleven are routine (billing questions, two refund requests, a feature ask) and all have drafts waiting for you.
Reads everything
Every ticket is read and classified the moment it lands, at 3am the same as 3pm.
Routes what matters
Urgent issues reach the right channel with the error, the accounts, and the context attached, not just a link.
Drafts the rest
Routine replies wait as drafts, so the human touch is a review, not a rewrite from a blank box.
Why the drafts are usable
A draft is only worth having if editing it is faster than writing it. What clears that bar is context: Cedric reads the ticket, the customer's history, and the systems that hold the answer, so the draft about a double charge contains the actual charges, in your team's voice.
And when the answer depends on a judgment call or a policy exception, the draft says so and asks. A confident wrong answer costs more than no draft at all.
Approval
Nothing reaches a customer without a yes.
CedricAPP10:02 AM
Reply drafted for the enterprise lockout: the SSO workaround, plus an offer of a call tomorrow. Send it?
Maya Chen10:03 AM
Move the call to Thursday and drop the last paragraph, then send.
CedricAPP10:04 AM
Sent. Thursday 14:00 offered, last paragraph removed. I noted the edit, so future lockout drafts will skip that paragraph too.
What the first week looks like
No migration, no implementation project, no new place for your team to check.
- 01
Add Cedric to the channel your support team already uses
One click. It joins your Slack like any other member, and everyone shares the same Cedric.
- 02
Connect the helpdesk, and nothing else yet
Point it at the one system that holds your tickets, with scoped and revocable access. Add billing and docs later, once the first loop is working.
Start narrower than feels useful. A Cedric that does triage well on day two beats one that does six things vaguely on day ten.
- 03
Ask it to triage, and read what it says
For the first few days, mention it each morning and check its classifications against your own judgment. Correct it in the thread when it is wrong; those corrections stick.
- 04
Turn on the drafts
Once the triage matches how your team thinks, let it draft replies. Everything still waits for approval, so the worst case is a draft you delete.
- 05
Schedule the digest
Volumes, response times, and the themes behind them, posted to the channel every Friday without anyone assembling it.
Four ways to run the queue
The honest comparison, including the two options that are genuinely fine for a lot of teams.
| By hand | Canned macros | Support chatbot | Cedric | |
|---|---|---|---|---|
| Reads overnight tickets | At 9am, by a person | No | Yes, front of queue only | As they land, all of them |
| Uses your live data | Yes, by looking it up | No | Rarely | Yes, from connected tools |
| Handles the unusual ticket | Yes, well | No | No, escalates it | Drafts it and flags the uncertainty |
| Customer sees unreviewed text | Never | Never | Always | Never, approval is required |
| Improves from corrections | Yes, per person | When someone edits the macro | Rarely | Yes, shared across the team |
| Cost of scaling | Another hire | Free, and it plateaus | Flat fee, deflection only | Credits per task |
What stays with your team
What goes to Cedric is the reading, the looking up, the first draft, and the remembering. What stays with your team is the judgment: the exception, the customer who deserves more than policy, the apology that has to be real.
Where this breaks
If your helpdesk is a shared inbox with no structure, triage is guesswork for a person and guesswork for Cedric. Fix the structure first; the tool cannot invent it.
If nobody reviews the drafts, quality drifts. Approval is not a formality that gets in the way, it is the mechanism that keeps drafts good, because the corrections are what teach it.
If your answers are not written down anywhere, Cedric will flag more uncertainty than you would like at the start. That is the right behaviour, and it is also a fair audit of your documentation.
Regulated replies still need your process. Cedric drafts and waits; it does not know which of your answers carry legal weight unless you tell it.
FAQ
Only if you set it up that way, and by default it does not. Cedric drafts, and a person approves before anything sends. Sending, changing, and spending always wait for a human yes.
From the replies your team has already sent, from any guidelines you give it, and from the edits you make to its drafts. Because the workspace memory is shared, a correction one person makes improves everyone's drafts, not just theirs.
Cedric connects to the tools your team already uses, with scoped permissions you grant per tool and can revoke at any time. If a system has an interface Cedric can reach, it can read the tickets there.
It drafts what it can, states plainly what it is unsure about, and asks. An unusual ticket surfaces as a question in the thread rather than a confident wrong answer to a customer.
Access is scoped per tool, logged, and revocable, and Cedric reads the context a task needs and nothing it does not. Nothing is connected until you connect it. See our [privacy policy](/privacy) for the detail.
Every workspace starts with $50 in credits, free, no card. Each task consumes credits in proportion to the work involved, so triage on a quiet night costs a fraction of a deep investigation. See [pricing](/pricing) for the plans.
It reads and drafts overnight, and it can post the urgent ones straight to the channel so the morning starts with the picture already assembled. What it will not do overnight is send anything to a customer unapproved.
Ready when you are
Put Cedric on the queue.
Install it where your support team talks and hand over the reading, the looking up, and the first draft.
$50 in free credits, no card needed.