Bring recurring inputs into one intake
Connect a folder, mailbox or supported remote storage source—or upload a document once. Changing file layouts feed the same draft contract.
- Set a source schedule and customer context.
- Discover the source’s documents and inspect each file status.
- Retain source identity, artifacts and content revisions rather than treating every fetch as a new forecast.
Follow durable processing and recover failed work
Discovery, preparation and extraction run as persisted jobs instead of holding a web request open.
- Inspect processed-file counts, active files and parallel requests.
- Cancel active runs or retry failed/cancelled work.
- Keep completed drafts and published history when another file fails.
Correct a draft with its evidence visible
Open a register record in place and correct the forecast lines before deciding whether to publish.
- Edit material, plant, period and whole-piece quantity; preserve an observed zero.
- Follow a validation issue to its field, then save and re-check.
- Retain source sheet/cell evidence; review nonblocking warnings separately.
Approve a version rather than overwrite history
Publishing is an explicit reviewer decision. The approved snapshot and its decision record remain inspectable.
- Lock and revalidate the submission before writing published lines.
- Retain rejected records and their reasons.
- Unlocking archives the publication before re-analysis; a changed source does not silently rewrite approved values.
Compare and deliver normalized planning data
Filter current forecast lines or retained snapshots, then inspect how two approved versions differ.
- Compare line-level additions, removals and changed values in an explicit A → B direction.
- Export the visible selection or the full filtered CSV/XLSX result.
- Deliver filtered output to a configured SQL destination; organization scope applies throughout.
Who does what
Configured intake and extraction run automatically in durable jobs. Review and publication remain human decisions.