Built to work with
the Financial Ombudsman.
The Financial Ombudsman Service exists to adjudicate genuine consumer grievances. AI is now being used to flood it with hallucinated complaints that waste caseworker time and dilute real claims. EvenStance does the opposite.
Every Frank letter is grounded in your specific facts, and citations are checked against legislation, with multi-agent review on paid tiers. You review it before it goes anywhere. That's what “built for FOS” means.
Start a complaint that holds upThe two-sided AI problem in financial services
AI is causing harm in two directions simultaneously, and EvenStance is the answer to both halves.
Regulators say frontier AI raises cyber risk
On 15th May 2026, the Bank of England, the FCA and HM Treasury said that “the cyber capabilities of current frontier AI models are already exceeding what a skilled practitioner could achieve”, and that regulated firms need to plan for and mitigate those risks in line with existing operational resilience rules.
Bank of England, FCA and HM Treasury joint statement on frontier AI models and cyber resilience, 15th May 2026
AI-generated slop is burdening the Financial Ombudsman Service
Simultaneously, AI is being used to flood the FOS with low-quality, hallucinated complaints. The FOS has highlighted the operational burden of AI submissions containing non-existent law references, incoherent arguments, and disproportionately long documents. Up to 35% of FOS responses to initial assessments may now be AI-influenced.
Financial Ombudsman Service, 2026
What AI slop looks like at the FOS
The FOS has been specific about what's going wrong. These are real problems affecting real complainants.
Hallucinated legislation
AI-generated complaints cite laws that don’t exist, case decisions that were never issued, and regulatory references that have no basis. FOS caseworkers have to verify all of it.
200-page responses to 6-page decisions
The FOS has noted AI is being used to generate “unduly lengthy or incoherent submissions”: in some cases 200-page responses to a simple 6-to-8-page provisional decision.
Up to 35% of sampled FOS responses may involve AI
The FOS estimated that AI contributed to up to 35% of responses to initial assessments in a sample analysis. Quality, not volume, is the issue for genuine complainants.
Why this matters for your claim: When FOS caseworkers spend their time fact-checking AI hallucinations in other complaints, that time isn't spent on your genuine claim. Well-prepared, factually grounded submissions move faster through the process. Slop doesn't just waste FOS time; it slows down everyone.
Frank's anti-slop architecture
Four design decisions that make Frank different from a generic AI prompt.
Reads your actual case data
Frank doesn’t hallucinate facts. It reads the specific details you’ve entered: the company name, dates, amounts, what happened, and what you’ve tried. Every letter is unique to your case.
Cites sector-specific legislation
Via EvenStance’s Sector database, Frank knows which legislation applies to your type of dispute (motor finance, energy, insurance, data protection) and cites it correctly, not generically.
Applies conviction scoring before escalating
Frank won’t recommend FOS escalation on a weak case. It assesses case readiness before recommending the next step. This protects the FOS from weak claims and protects you from wasted effort.
You review it before it goes anywhere
Every Frank-generated letter is reviewed by you before sending. No AI sends anything on your behalf without your explicit approval. You are the decision-maker; Frank is your drafter.
What a FOS-ready complaint actually looks like
A typical AI-written complaint
- • Cites “FCA Regulation 2019/42” (doesn't exist)
- • References a fictional ombudsman decision from 2023
- • 47 pages of generic consumer rights boilerplate
- • No specific dates, amounts, or evidence referenced
- • Demands “full compensation and punitive damages” (not available at FOS)
A complaint drafted with Frank
- • Cites the correct Act, section, and why it applies to your specific facts
- • References your actual dates, agreement amounts, and company responses
- • 2 pages: clear, structured, and evidence-led
- • Specific remedy requested with the legal basis for it
- • Realistic expectations: what FOS can and cannot award
Start a complaint that will be taken seriously
Frank generates complaints grounded in your facts, citing real legislation, reviewed by you before anything is sent. That's what the FOS needs more of.
Start with Frank