In fiscal year 2025, U.S. federal agencies received 1,707,197 Freedom of Information Act requests — a record, and 13.7% more than the year before. The Defense Department's backlog alone grew 42% to over 30,000 unresolved cases. Most agencies responded to that surge the same way they have since 1967: by adding people to a process built for paper.
Throwing headcount at a spreadsheet-driven intake queue doesn't scale, and hiring freezes across federal FOIA offices have made it worse. There's a narrower fix that doesn't require a bigger office: let AI handle triage, drafting guidance, and duplicate detection before a human ever opens the request.
A major U.S. defense agency has been running exactly that experiment since 2021, using Casepoint's eDiscovery and FOIA platform to replace manual review and spreadsheet-tracked legal holds. The agency now consistently meets statutory response deadlines and has cut its backlog significantly — proof that the bottleneck was never staffing, it was workflow.
What Casepoint's AI Assistants Actually Automate
In May 2026, Casepoint shipped three purpose-built AI assistants inside its existing FOIA platform. The AI Reading Room Assistant scans a requester's query against records already published and redirects them to existing disclosures before a new case file is even opened — cutting volume at the source. The AI Request Writing Assistant (patent pending) helps requesters scope vague submissions into something an analyst can actually action, reducing the back-and-forth clarification emails that eat weeks off every timeline.
The third piece, the AI Intake Assistant, is where most of the time savings live: it automatically flags incoming requests for completeness, detects near-duplicate submissions from the same requester or organized campaigns, and screens for bot-generated volume — the kind of automated flooding that has quietly become a real problem for understaffed FOIA offices. Casepoint's own analytics show purpose-built workflows like these helping small teams process massive volumes up to 75% faster.
If you run a FOIA, records-management, or open-records office — federal, state, or municipal — this is built for you. It's equally relevant to general counsel teams handling public-records litigation risk, since faster, better-documented responses reduce exposure to failure-to-respond lawsuits.
Why the Backlog Keeps Growing Even With More Staff
The instinct to solve a request-volume problem by hiring more analysts assumes the bottleneck is reading and redacting documents. In practice, a large share of analyst time goes to work that has nothing to do with substantive review: figuring out whether a request duplicates one already in the queue, chasing a requester for clarification on an ambiguously worded ask, or manually checking whether the answer already sits in a published reading room.
Casepoint's approach targets exactly that pre-review layer. The State Department's FOIA backlog grew by 6,000 cases in a single year even as staff worked overtime — a strong signal that the constraint isn't willingness to work, it's the volume of low-value triage sitting in front of every real review.
Agencies that automate this layer now are also building a defensible audit trail earlier — every AI-assisted triage decision is logged, which matters when Congress or the courts ask an agency to justify its FOIA processing timelines. Offices still running intake by email and shared spreadsheet are, in effect, one workforce reduction away from a much worse FY2027 backlog report.
The Results Agencies Are Reporting
Casepoint's platform is trusted by more than 150 federal, defense, and state/local/education (SLED) customers, and the company has processed FOIA and eDiscovery matters under FedRAMP High, DoD Impact Level 5, and DoD Impact Level 6 authorizations — Casepoint is one of only six providers worldwide with that IL6 distinction, a relevant detail for any agency handling controlled unclassified information alongside FOIA requests.
The defense agency case study referenced above is now in its fifth year on the platform, with FOIA sitting alongside litigation holds and investigations on the same system — meaning records pulled for one legal process don't need to be re-collected for another.
Agencies that modernized their FOIA intake in 2025-2026 are already citing shorter time-to-first-response in their own Chief FOIA Officer reports. The ones still triaging by inbox are the agencies showing up in next year's "largest backlog increase" headlines.
Context matters here: government-wide, agencies processed a record 1,635,055 FOIA requests in FY2025 alongside that record 1,707,197 received — proof the system as a whole is barely keeping pace even before accounting for individual agencies like Defense, State, Education, and Transportation reporting double-digit backlog growth in the same period. A platform change that shifts even the pre-review triage layer compounds across every request an office touches, not just the ones that make headlines.
The Platform Behind the Case Study
This is enterprise government software — pricing is quote-based, not published, and evaluation typically runs through a formal procurement process rather than a self-serve trial.
| Tool | Role in This Workflow | Free Tier? | Paid From |
|---|---|---|---|
| Casepoint FOIA | Core platform: request intake, review, redaction, production tracking | No | Quote-based (government contract) |
| AI Reading Room Assistant | Deflects requests already answered by published records | Included | — |
| AI Request Writing Assistant | Helps requesters scope ambiguous submissions | Included | — |
| AI Intake Assistant | Flags incomplete, duplicate, or bot-generated requests | Included | — |
⚠️ [REQUIERE VERIFICACIÓN] Exact per-seat or per-request pricing is not publicly disclosed; agencies should request a scoped quote directly from Casepoint through FedRAMP marketplace channels.
Who This Case Study Is For
This is most directly relevant to FOIA officers, records-management directors, and general counsel at federal agencies, state governments, and larger municipalities already fielding more than a handful of requests per week. It's also useful reading for agency CIOs evaluating whether their next eDiscovery or litigation-hold platform should also cover FOIA, since running both on one system avoids duplicate document collection. It's not the right fit for a small local office handling a handful of requests a year — at that volume, a lighter-weight records-request tracker will do the job without a formal procurement cycle.
Frequently Asked Questions
Does an AI assistant decide what gets released under FOIA?
No. Casepoint's AI assistants handle triage, drafting help, and duplicate/completeness screening — every release, redaction, and exemption determination is still made by a human FOIA officer. The AI narrows what reaches a reviewer's desk; it doesn't replace the reviewer.
Is this only for federal agencies?
No — Casepoint also serves state, local, and education (SLED) customers, and the same triage/deflection logic applies to any public body fielding open-records requests, including agencies operating under state-level transparency laws rather than the federal FOIA statute.
How long does a FOIA platform migration take?
For an agency already running structured intake, initial deployment of the AI assistants typically layers onto an existing Casepoint implementation rather than requiring a rebuild — the defense agency case study above started with core eDiscovery in 2021 and expanded from there, rather than a single big-bang FOIA rollout.
FOIA request volume has grown every year this decade, and headcount hasn't kept pace with it in a single one of them. The agencies narrowing that gap aren't the ones asking for bigger budgets — they're the ones automating the triage nobody wanted to do manually in the first place.