The missing items list is the most valuable page in the pack
To fulfill an audit request, you have to identify documents for a given period and a specific set of document types, searching across a document store, an email archive, an accounting system, and someone's laptop.
Collecting is boring, and because of that it is done at the last moment, which is when the gaps are found.
A pack with an honest missing-items list is accepted more readily than one that arrives late but complete. If the gap is known, the auditor can plan around it. If the document is missing when week three arrives, they cannot.
Indexing is what makes a pack more useful than a folder. If an auditor can find the invoice referenced in a sample line within 10 seconds, they ask fewer follow-up questions. Those follow-up questions are the real cost of an audit.
How the ActionFlow runs on the canvas
The audit request names the requirements. A Google Sheets or Airtable node stores that list as structured rows: document type, period, entity, and the usual location for each type.
A Start node begins the run. A Loop node walks the requirement list and reads from the right source each time: an S3 or Dropbox node for stored files, a PostgreSQL node for ledger extracts, an Email node for correspondence, and a Notion node for policies.
A Generate Object node compares what was found with what was requested and returns three arrays: satisfied, partially satisfied, and missing. Partial satisfaction is a real state. Collapsing it into a neighbor leads to surprises later.
A Validate node confirms that each document found actually covers the requested period, rather than merely existing.
A Create Excel node builds the index: requirement, document, and location. An S3 or Dropbox node gathers the files into one structured folder.
A Slack node posts the missing list to the finance lead before any pack is sent, because the missing list is a task, not a report.
Nothing goes to the auditor without approval from a Human in the Loop node.
Nodes this ActionFlow uses
- Airtable (Integration): holds the requirement list with normal storage location per type. Google Sheets fits the same slot.
- Loop (Control): walks the requirement list one item at a time.
- S3 (Storage): reads stored documents. Dropbox and FTP fit the same slot.
- PostgreSQL (Data): supplies ledger extracts for the period.
- Generate Object (AI Core): returns satisfied, partially satisfied, and missing as separate arrays.
- Validate (Data): confirms each document covers the requested period, not merely exists.
- Create Excel (Data): builds the index that makes the pack navigable.
- Slack (Human in the Loop): delivers the missing list first, and gates anything sent to the auditor.
Frequently asked questions
Start building AI workflows
Create a free account, open a template or a blank canvas, and run your first ActionFlow.
Newsletter
Get product updates
New nodes, agents, and product notes. We send mail only when we have something worth opening.