Use case · Operations
The glue work, done.
Ops is the department of everything that falls between departments. Cedric picks up the recurring, cross-tool work that has no natural owner.
$50 in free credits when you sign up. No card needed.
- Operations work is defined by the fact that it crosses tools and teams, which is exactly why it resists automation.
- Cedric works across the systems you connect, so a task that spans four tools is one request rather than four.
- The recurring checks (access reviews, vendor renewals, ticket ageing) run on a schedule and report by exception.
- Chasing becomes a scheduled nudge rather than a person remembering to be annoying.
- Anything that changes access, spends money, or messages people waits for approval.
The work that has no home
Every company has work that belongs to nobody and stops everything when it lapses: who still has production access, which vendor renews in three weeks, whether the new starter has accounts on day one.
It resists ordinary automation because it spans systems. The task is usually "compare this over here to that over there and tell me what does not line up", which used to mean a person with several tabs open.
Recurring checks
The access review that actually happens.
Yara Nasser9:14 AM
@Cedric audit who still has prod access
CedricAPP9:15 AM
14 accounts with production access. Five look wrong: two people left the company (last active 41 and 63 days ago), one moved to marketing in March, and two are service accounts nobody has touched in six months. The other nine are current engineers with a reason to be there. List attached with last-activity dates. Want me to draft the removal requests for the five?
By exception
You get the five that need attention, not a spreadsheet of fourteen for you to read through.
Across systems
It compares the access list against the people directory and activity data, which is the comparison that makes it meaningful.
On a schedule
Set it monthly and the review stops depending on someone remembering it exists.
Chasing, without being the person who chases
A surprising share of operations work is reminding people about things they agreed to do, and everyone resents whoever does it. From Cedric, the nudge is consistent, on time, and not from a colleague who now feels like a nag.
It also carries the context: what the item is, why it blocks, how long it has been open. That is the difference between a reminder that gets actioned and one that gets ignored.
Onboarding
Day one, ready before day one.
Felix Andersson10:02 AM
@Cedric prep the onboarding for Monday's hire
CedricAPP10:03 AM
Prepped for Priya, starting Monday on the platform team. Accounts requested for the six tools that role gets, with the two that need manager approval flagged to you. Buddy assigned (Kevin, based on team and timezone). First-week calendar drafted with the three standing meetings and two intro calls. Laptop shipping confirmation is missing, that is the only open item and it is with IT.
What to hand over first
Ops has the widest surface of any function, so the trick is picking a narrow start.
- 01
Pick the check you keep forgetting
Whichever recurring review has lapsed most often is the best first candidate, because the bar is currently zero.
- 02
Connect the two systems it compares
Most useful ops checks are a comparison between two sources of truth. Connect both, scoped and revocable.
If the check only needs one system, a simple alert probably already solves it. The value here is in the crossing.
- 03
Run it manually until you trust the exceptions
Ask for it in the channel a few times and check the flagged items yourself. You are calibrating what it considers an exception.
- 04
Put it on a schedule
Then it reports by exception on its own cadence and you only hear from it when something needs you.
- 05
Add the chasing
Once a check reliably identifies who is blocking what, let Cedric do the reminding. Messages to people still wait for your approval.
Compared to the usual approaches
Ops teams have tried all of these, usually at once.
| A checklist doc | Workflow automation | A dedicated ops hire | Cedric | |
|---|---|---|---|---|
| Runs without being remembered | No | Yes, on triggers | Yes | Yes, on a schedule |
| Spans several systems | A person does the spanning | With integrations built per flow | Yes | Yes, natively |
| Handles an unexpected case | The person improvises | It breaks or does nothing | Yes | Flags it and asks |
| Changing the process | Edit the doc, hope people read it | Rebuild the flow | A conversation | A conversation, in the channel |
| Reports by exception | No | Depends on the build | Yes | Yes, by default |
| Changes access unattended | No | If configured to | Yes | Never, approval required |
Where this breaks
If the two systems disagree about identity, comparisons get noisy. When someone is `j.smith` in one tool and `jsmith@` in another, expect false positives until the mapping is sorted. Cedric will say it is unsure and leave the call to you.
A process nobody agreed on cannot be automated into existence. If onboarding is different every time because nobody decided what it is, Cedric will surface the inconsistency, which is useful, but the decision is still yours.
Access changes stay gated. It can find the five accounts that should not have production access and draft the removals. Actually removing them waits for a person, every time.
It is not a compliance programme. Evidence collection and recurring reviews help, but attestations and control ownership are human responsibilities with names attached.
FAQ
It can prepare the change and tell you exactly what it will do, then it waits. Anything that changes access, sends, or spends requires an explicit human approval.
Those are excellent when the flow is known in advance and stable. Cedric handles the cases where the request is phrased in words, the shape varies, or the answer requires judgment about what counts as an exception. Most ops teams end up with both.
Yes, that is the common case. It connects to each system separately with scoped access and does the comparison itself, so the tools never need to know about one another.
It tells you what needs attention and stays quiet about what does not. A monthly access review that finds nothing wrong posts one line saying so. You should not have to read a report to find out that nothing happened.
Anyone in the channel, which is often the point: the person who notices the problem can just ask, with no ticket to file and no queue to wait in.
Yes, and it is the recommended start. Checks and reports need no write access at all, and they are usually where most of the value is.
Ready when you are
Give the glue work an owner.
Install Cedric where your ops team talks and hand it the check you keep forgetting to run.
$50 in free credits, no card needed.