Help that has already read your ledger.
It answers from your records, not from a manual.
Ask why an invoice will not post, where a delivery went, or what changed on an order, and the answer comes back citing your own records — not a help article about the general case.
Not a manual. Your actual records.
Ask why an invoice will not post, where a delivery went, or what changed on an order, and the answer comes back citing your own records. When it needs a person, it hands over the whole trail rather than asking you to repeat yourself.
The answer comes from your records, not a help article
The answer is assembled from your records, under your permissions, and reproduced on a copy before it is given. Six seconds, seven records read, all seven cited.
- You ask
- It reads
- It reproduces
- Answer
- Question
- “Why on hold?”
- About
- SO-10407
- Asked by
- D. Ashworth
- Scope
- Your permissions
- Records
- 7
- Cited
- All 7
- Ran on
- A copy
- Live data
- Untouched
- Result
- Reproduced
- Reason
- Credit check
- Evidence
- Linked
- Escalated
- No
Answers about your system
“Why is this order still on hold?” gets the reason from that order — the credit check, the missing date, the line that will not allocate — not a help article about holds in general.
It reproduces before it replies
For anything it cannot answer from the record, it re-runs the step on a copy and shows what happened, so the answer is what your system did rather than what it should do.
Handover keeps the context
When a person is needed, they arrive with the records, the steps taken and what has already been ruled out. You are not asked to explain it twice.
It writes down what it learns
Anything resolved once becomes an answer the next person gets straight away, described in your own terminology rather than ours.
Reading, not acting.
Support answers and explains. Anything that would change a record goes through the same approval as everything else — the support agent has no privileged path to your data.
- It reads under your permissions — it cannot see records the person asking could not see.
- It cannot change anything on its own — a fix it proposes is prepared and queued for approval like any other agent action.
- Reproduction runs on a copy — never on your live data.
- Every answer cites its sources — you can open the records it read from the answer itself.
- Escalation is a handover, not a restart — the person who picks it up sees the whole trail.
Related pages.
Frequently asked questions.
Is this a chatbot on the docs?
No. It reads your actual records. “Why is this order on hold?” is answered from that order — the credit check, the missing date, the line that will not allocate — with the records cited.
Can the support agent change my data?
No. It reads and explains. Anything that would change a record is prepared and queued for the same approval as any other agent action.
What happens when it cannot answer?
It hands over to a person with the records, the steps already taken and what has been ruled out. You are not asked to explain the problem a second time.
Is support included in my plan?
Every plan includes support — email on Launch, email and chat on Growth, priority from Scale upward. See pricing. AI support is an optional add-on alongside it.
Stop explaining your setup to support.
The answers come from your records. When a person is needed, they already have the trail.
Join waitlist