How NoReply works

When you start a complaint, your situation passes through a sequence of stages. The aim is the same every time: turn a free-form description of what went wrong into a focused letter that cites the right law and lands in front of the right person at the company.

First, intake. You describe what happened in your own words and attach anything that backs it up - receipts, screenshots, photos, recordings. There's no template to fight. The richer the description, the more accurate the analysis.

Second, analysis. The system reads your account, identifies the most relevant consumer-law angle (refund right, Section 75, UK261, GDPR subject access, Consumer Rights Act fault claim, and so on), and judges how strong the case is. Evidence files are processed too, so a blurry receipt and a half-remembered phone call still contribute facts.

Third, research. We pull the actual current contact route for the company - escalation team, complaints inbox, ombudsman fallback - plus the specific statute text the letter will quote. The statute is the real text, not a paraphrase.

Fourth, draft. A complaint letter is composed that opens with the facts, cites the relevant law verbatim, states what you want, and gives the company a clear deadline. You review it, edit anything you want, and send it. NoReply never sends on your behalf without your explicit click.

Generative AI is the right tool for this job for a few specific reasons: it can read messy unstructured text and pull out the legally-relevant facts, it can match those facts to the right statute, and it can write a letter that reads like a person rather than a form-filling exercise. None of those individually are new ideas; doing all three reliably in the same pipeline is what makes the experience feel different to a template generator.

Where we apply AI deliberately: case classification, statute matching, multimodal evidence reading (images, PDFs, audio), and the actual draft. Where we don't: anything where a wrong answer could mis-cite the law. Legal citations are pulled from a curated set of primary sources, not generated by the model.

We keep the model + version used on every generation, so any letter ever produced can be reproduced exactly. If a citation is wrong, we can identify which model run produced it and fix the underlying prompt or source.

We cite primary legislation, not summaries. The most-used statutes:

UK261 (Civil Aviation (Denied Boarding, Compensation and Assistance) Regulations 2005) for flight delays and cancellations on UK-departing flights or UK-airline arrivals. Compensation amounts mirror the post-Brexit UK schedule: up to £520 depending on delay length and flight distance.

EU261 (Regulation (EC) No 261/2004) for EU-departing flights and EU-airline arrivals into the EU. Up to €600 on the same delay/distance basis.

Consumer Rights Act 2015 for goods and services bought in the UK. Sections 9-18 (goods of satisfactory quality), 19-24 (remedies for faulty goods), 49-56 (services), 62-63 (unfair terms).

Section 75 of the Consumer Credit Act 1974 for credit-card purchases between £100 and £30,000 where the supplier fails to deliver or the goods are faulty.

UK GDPR / Data Protection Act 2018 for subject access requests (Article 15) and rectification/erasure (Articles 16-17).

Package Travel and Linked Travel Arrangements Regulations 2018 for package-holiday issues.

Where law differs by jurisdiction (US DOT for US-domestic flights, Canada APPR, EU EC261), the letter cites the rule that applies to your specific case, not all of them.

Live flight disruption data is refreshed throughout the day from public flight-status sources. We rotate between sources so a single one going down doesn't take the board with it.

The UK Consumer Complaint Index pulls quarterly regulator data: Ofcom (telecoms), Ofgem (energy), and CAA (aviation). Each is published under Open Government Licence v3.0. The full data-science methodology - which complaint categories we count, how we weight them, how rolling averages are calculated - is published separately at /data/methodology.

Company contact information is a curated index, built and maintained by us. It's the part of the product that's hardest to fake; companies move their complaints addresses, escalation teams rotate, ombudsman jurisdictions change. We keep up to date.

We don't buy data from third-party brokers, and we don't enrich user records from external sources. The only personal data we hold about you is what you give us directly.

NoReply is not a law firm. We generate letters that cite UK consumer law and explain your rights, but we do not provide legal advice, represent you in court, or draft court documents. For anything that needs a solicitor, see Citizens Advice or contact one directly.

We don't take a cut of any compensation you recover. Unlike claims management companies, you keep 100% of what you win. Our revenue comes from subscription plans and pay-per-case credits, not from your settlement.

We don't send anything to a company without your explicit confirmation. The AI drafts; you review; you send. No background autopilot.

We don't use your complaint text or evidence to train models. Conversations with the AI are not retained for training purposes by us, and we operate under data-processing agreements with each AI provider that prohibit them using inputs to train their public models.

We don't promise you'll win. Some complaints are weak even when written well; some companies refuse anything short of a court date. The strength score on each case is honest about that. The letter still gets you in front of the right team with the right legal framing.

We chose accuracy over speed. The full pipeline takes longer than a template generator would. The trade is that the letter is genuinely specific to your situation rather than a fill-in-the-blanks form.

We chose primary sources over summaries. A draft cites "Consumer Rights Act 2015, s.20(7)" rather than "the Consumer Rights Act" because the company's complaints team needs to look it up. Generic citations get ignored.

We chose explicit user confirmation over autopilot. It's slower than a one-tap "go" button, but it keeps the letter yours and means you control what's actually sent. That's a feature, not a bug.

We chose UK-first depth over global breadth. We support US and Canadian regulations where they apply to your case, but the deepest coverage is UK consumer law because that's where most of our users are.

Every letter you generate is fully editable before sending - and after sending, the citation text is retained alongside the case so you can verify it against the legislation we name.

The legal citations in your letter link to legislation.gov.uk where applicable, so you can read the actual statute we relied on.

If you ever spot a citation that's wrong, email hello@usenoreply.com with the case ID. We treat that as a P0 bug: a wrong citation undermines the whole product.

For transparency on system reliability and recent activity, see the live system status at /status.