AI-Generated FOI Request Triage & Casework Guardrails

Information-governance teams must still apply FOIA law request by request. Higher volumes and more clarification work consume scarce officer time, increase deadline risk and make it harder to distinguish genuinely complex requests from machine-generated noise.

The buyer is usually an information-governance, legal or digital-services lead. Users need assistance that remains explicitly human-controlled because an automated tool cannot determine legal validity or exemptions on its own.

The problem is not that authorities lack case-management databases; it is that existing workflows were designed before requesters could cheaply generate long, legally styled correspondence at scale. That creates a new triage and quality-control requirement.

Interview 10–15 FOI managers in councils/NHS bodies; measure monthly AI-suspected request volume, clarification minutes per request, duplicated/repetitive requests and missed-deadline risk. Pilot as an add-on to existing case systems rather than replacing them.

A defensible wedge would be AI-specific request diagnostics with auditable explanations, configurable local policy, previous-response retrieval and integrations into incumbent FOI systems. Avoid claiming that AI can decide whether a request is valid, vexatious or exempt.

Reduce avoidable clarification work and administrative handling time while preserving lawful, transparent human decision-making.

Directly evidenced — medium confidence.

Pilot £3,000–£8,000 per organisation for 8–12 weeks, then roughly £4,000–£15,000/year depending on request volume and integrations. Benchmark: Workpro Requests is £480/user/year on G-Cloud, so an AI triage add-on must show measurable officer-time savings rather than simply duplicate case management.

Evidence basis: Workpro FOI (£480 per user per year) is the closest verified direct anchor used here. Its buyer, duration and scope are not assumed to be identical; implementation is separated where the opportunity requires integration, assurance or managed delivery.

Ask one accountable compliance, operations, legal or assurance owner to fund a paid test of AI-Generated FOI Request Triage & Casework Guardrails lasting 8–12 weeks, using an opening price of £3,000–£8,000 per organisation and covering 20 live cases, checks, submissions or evidence packs from one controlled workflow. Paid scope: A human-in-the-loop FOI triage layer that analyses incoming requests for clarification needs, likely duplicate themes, questionable statutory references and complexity, then prepares an officer-review checklist. Charge by organisation, site, user or completed case/check and compare the fee with manual review, external-assurance and evidence-chasing effort. Measure evidence completeness, review time, exception accuracy, rework, overdue actions and accepted submissions. Continue only if handling/rework falls by at least 25%, at least 90% of required evidence is complete and no critical exception is missed. Stop or reprice if false assurance creates a material miss, users bypass the workflow or saved effort does not justify the fee.

Raise the score if pilots show a material reduction in clarification time or repeat-request handling and incumbent integrations are feasible. Lower it if FOI vendors rapidly ship equivalent functionality or authorities report that AI-generated requests are still a small share of workload.