How the approval gates, the audit trail and the pricing logic actually behave.
No. Every agent output is a draft. A user with the Approver role has to review and release it before anything leaves the system — that gate cannot be turned off.
The field locks. Agents stop rewriting any value a human has touched, so your correction survives every subsequent run.
Every mutation is written to a hash-chained audit log recording whether a human or a specific agent run made the change, with a correlation ID tying it back to the originating request. Quotes are also snapshotted per version.
Prices outside their validity window are excluded from costing rather than silently used. If a line has no usable cost, the run parks and produces a partial quote flagging the gap instead of inventing a number.
Yes. Quotes are drafted in the customer's own currency using the exchange rate held in the system, and margin is computed on the landed cost.
Every row carries an organisation ID and isolation is enforced in the database itself, not only in application code.
Yes. Workflows are durable checkpointed graphs, so a run waiting on your approval resumes from where it stopped.
On-premise and private-cloud deployment are available on the Enterprise plan.
Most teams import their product master, customers and vendor price lists and are drafting real quotes in the first week.
Still unsure? Talk to us →