Agentic flooding is real. Your intake queue is next.
A new study documents agentic flooding across 84 cases in 11 jurisdictions as AI writes complaints and claims. What cheap filing costs do to your support queue.
The cost of writing a formal complaint just went to roughly zero, and the queues that receive those complaints are buckling. A paper from Chris Schmitz, Lewis Hammond of the Cooperative AI Foundation, and Alan Chan of GovAI documents agentic flooding — surges in filings to public services driven by large language models generating text cheaply. TechCrunch reported on the research this week. If you run a business with any intake channel — support, warranty, refunds, disputes, quotes — this is a preview of your next twelve months.
What actually happened
The paper, Characterizing Agentic Flooding of Government Services, examines 84 potential flooding cases across 11 jurisdictions and concludes the phenomenon is likely happening widely today. The numbers TechCrunch cites are not subtle. Complaints to the UK housing ombudsman rose from 2,600 in 2022 to just over 7,000 last year. The US Consumer Financial Protection Bureau saw roughly 5x growth in complaints across the same period. The work is due to be presented at the AI Ethics and Society conference next month.
The finding that matters is who is doing the filing. This is not a spam story. "The vast majority of cases we find are people who are entitled to claim for something, claiming for that thing," Schmitz told TechCrunch. The barrier was never eligibility. It was the four hours of writing, the formatting, the knowing-which-form. An LLM removed that, and the latent demand showed up all at once.
The authors also flag the obvious wrong fix: agencies can throttle the flood by making filing expensive again, but cost-based barriers hit the people who needed the help most.
Why agentic flooding matters for your business
Your customers have the same tools. The person who would have grumbled and moved on now has a well-structured, citation-laden, three-page dispute letter in ninety seconds — and a follow-up sequence if you do not answer.
The wrong response is the one the agencies are being warned about: add friction. Longer forms, phone-only returns, a portal nobody can log into. That filters your good customers faster than your bad ones, and it is the exact posture that gets a business posted about.
The right response is to make the volume cheap to serve. Three moves, in order:
Structure the intake. Free-text email is the most expensive input you can accept. A short structured form — order number, category, desired outcome, evidence upload — turns a three-page letter into five fields your system can route without a human reading prose.
Auto-resolve the deterministic tail. A meaningful slice of any support queue is requests with one correct answer: order status, a refund inside policy, a reship on a confirmed-lost package. Those should never touch a person. Wire the rules to the systems of record and let them close themselves.
Route the rest with the context attached. When a human does get involved, they should open a ticket that already has the order history, the prior contacts, and the policy citation on screen. That is the difference between four minutes and twenty.
Capacity planning matters too. If your queue doubles, the question is whether that costs you two headcount or nothing. That number is decided by how much of the work is structured today, not by how fast you hire later.
Key takeaways
- A new paper by Schmitz, Hammond, and Chan documents agentic flooding across 84 cases in 11 jurisdictions
- UK housing ombudsman complaints went from 2,600 in 2022 to over 7,000 last year; the CFPB saw roughly 5x growth
- Most of the surge comes from legitimate claimants who previously found the process too burdensome, not from spam
- Adding friction to slow the volume punishes your best customers first — do not copy that fix
- Structure your intake, auto-resolve the deterministic tail, and attach context before a human ever opens the ticket
- Whether a doubled queue costs you two hires or nothing is decided by how much is structured before it doubles
If your support queue doubled next quarter, would you notice in the metrics or in the payroll? We build structured intake, rules-based auto-resolution, and context-loaded routing on top of the systems you already run — so volume stops converting directly into headcount. Run the numbers on your ticket volume, or send us your current intake flow and we'll tell you what's automatable.
Sources: TechCrunch, arXiv.
- #ai-agents
- #customer-support
- #intake-automation
- #operations
- #workflow-design
Tommy Rush — Founder, Rush Commerce
Operator turned builder. 15+ years running operations — now shipping the systems businesses run on. More
Get The Rush Report weekly — one email, zero fluff.
Keep reading
Senate AI duty of care draft: model releases get a veto
Thune, Cruz and Klobuchar are drafting a federal AI duty of care with power to block unsafe model releases and preempt state law. What it means for your stack.
Read itOpenAI won't IPO in 2026: your core vendor stays private
Sam Altman says an OpenAI IPO in 2026 would be 'ill-advised' given safety concerns. What a private core AI vendor means for the small businesses built on its API.
Read it