ezInputsezInputsDocs

Product

Submissions & workflows

Every submission lands in a tenant-scoped review queue with a full audit trail. AI can soft-approve in bulk — but only humans change status.

The review panel — submission answers on the left, AI recommendation and decision controls on the right.
The review panel — submission answers on the left, AI recommendation and decision controls on the right.

Workflow lifecycle

Submitted

Hits the queue

Pending

Awaiting review

AI signal

aiReview (advisory)

Human

Approved or rejected

Status set

pendingneedsInfoapprovedrejected+ custom workflow steps

AI soft-approval signals

highlights[]

Things the AI noticed are good — quoted from the submission.

concerns[]

Things worth a closer look. Each links back to the source field.

fieldsReferenced[]

Field IDs the AI quoted — the UI resolves these to labels.

suggestedAction

Recommended next step: approve, reject, or needs-info.

What AI does

Writes the aiReview advisory field. Annotates concerns and highlights. Surfaces a suggested next action.

What AI never does

Change workflow status without an HMAC confirmation token. Bulk-approve > 25 records without explicit human confirm.

Triaging via Claude

"Show me pending submissions on the Event Feedback form."
"Recommend approval for any pending where 'satisfaction' is 4 or 5 and no complaints."
"Use the /triage-submissions prompt on the Volunteer Sign-up form."

See MCP overview for the full submission tool catalog and audit-log details.