Grant applications processed on one record

The problem

The application arrived through the public portal, but everything after that lived somewhere else — eligibility notes in a spreadsheet, reviewer comments in email threads, disbursement status in a legacy case system nobody outside finance could see. Reconstructing the history of a single award for an audit meant piecing it back together from three systems and whoever remembered the context.

What we configured

Four capabilities on the same platform foundation, no bespoke build:

Intake

Applications, budgets, and supporting documents are captured from the existing portal, no new submission system for applicants to learn.

Eligibility screening

Automated checks against programme rules flag ineligible or incomplete applications before a reviewer opens the file.

Scoring support

Reviewers see a structured summary and a consistency check against scoring criteria, not just a raw document stack.

Disbursement tracking

Award, payment schedule, and reporting obligations live on the same record as the original application.

Where humans stayed

Eligibility and completeness get checked automatically; the funding decision does not. Every score, every note, and every override a reviewer makes is attached to the record, so the audit trail reflects an actual human decision rather than a system default.

[Client quote pending approval]

— [name, title]

What came next

The same intake and audit-trail foundation is now being configured for a second grant programme in the same office, reusing the eligibility engine with a new rule set rather than a new build.

Sources

One real-world use case of AI in government, based on: