Help center

Set up, connected, running.

Everything about getting Clarhet reading your plan and your systems. If the answer is not here, a person replies to hello@clarhet.ai.

01

Getting started

What do I need before signing up?
A business email and your current plan in whatever form it exists: a doc, a sheet, a deck, or a set of quarterly goals in a tool. Clarhet reads it as written and asks about the parts that are ambiguous.
How long until the first useful read?
Most teams get a first morning read within a few days of connecting two systems. It gets sharper as more sources come in, because the agents can then check one against another.
Do we have to change how we work?
No. Clarhet derives work from the plan and the systems of record you already use. Nobody is asked to keep a second list up to date.
02

Sources and access

Which systems can Clarhet read?
Warehouses, CRMs and ERPs, spreadsheets, operational and support tools, documents, internal APIs, and vendor portals that sit behind a login with no clean API. Public and market data comes from the open web.
What permissions do you need?
Read access, scoped to the sources you nominate. Where writing is useful, such as drafting an update or closing an action, it happens in a system only after you turn that on for it.
Can we disconnect a source later?
Yes, at any time. The read rebuilds without it, and figures that depended on it are marked as unverifiable rather than estimated.
03

The daily read

How is my plate decided?
Items are derived from the plan and from movement in your systems, then filtered to what needs your decision, your signature, or your check on something already drafted.
Why does a number look different from my dashboard?
Open the figure and you will see every source behind it and how conflicts were reconciled. Usually the difference is a definition, and the trail shows exactly where it enters.
What happens to items I never action?
They stay on the record with their age visible, and the read tells you what is waiting on you rather than quietly dropping it.
04

Agents and autonomy

What do the agents do without asking?
They read, reconcile, draft, and chase. Anything that commits you to something outside the company waits for a person, and every action is written to a log you can review.
Can I widen or narrow what they handle?
Yes, per source and per kind of task. Most teams start narrow, read the log for a week, then hand over more.
What if an agent gets something wrong?
Correct it on the item and the correction is kept with the trail. If a figure cannot be verified against a source, it is not shipped in the read at all.
05

Team and roles

Who should be in the account first?
Whoever owns the plan, plus one person per function that owns commitments in it. That is enough for the read to be real without a rollout.
Does everyone see everything?
No. Permissions are scoped to your team and your sources, so people see the slice they act on and the figures they are cleared for.
06

Billing

How does billing work?
Per seat, monthly or annual, on the plan you pick at sign-up. Annual terms are billed once for the year. See pricing for current rates.
Can we change plans or seat count?
Yes. Changes take effect on the next billing date, and added seats are prorated for the current period.
How do we cancel?
From the account settings, or by writing to us. Access runs to the end of the paid period, and you can export what the agents assembled before it ends.

Still stuck? A person answers.

We are small enough that support reaches the people who built it. Send the question with your account email and we will pick it up.