AI-powered dispute resolution for English law.
Enter the password to continue.
ResolveThis is an online dispute resolution platform for English law. Disputes are resolved by AI trained by practising barristers and retired judges, with the option of human review by a named lawyer or judge. The outcome is a formal arbitral award, enforceable under section 66 of the Arbitration Act 1996.
A civil claim in the County Court takes over a year to reach trial. Solicitors' fees run to tens of thousands of pounds. Most individuals and smaller businesses cannot afford it. So legitimate claims go unresolved. Or defendants have to pay over the odds on a 'commercial' basis because the cost of lawyers is so high.
The problem is structural. English litigation was not designed for the volume or cost sensitivity of modern disputes. Most of the cases that should be brought never are. That is the market ResolveThis is built to serve.
The question is not whether most disputes will be resolved by AI. They will. The question is who builds the platform.
| County Court (HMCTS) | ResolveThis | |
|---|---|---|
| Time to resolution | Small claims track: median 39 weeks to trial. Fast/multi-track: median 61 weeks to trial. | 3 to 5 weeks from submission to binding award. |
| Court or platform fees | £35 to £455 issue fee (claims up to £10,000). 5% of claim value for higher claims, capped at £10,000. Hearing fee on top. | Usage-based. Approximately 20p per page, 20p per minute of interview, 50p per image or video minute. Typically low hundreds of pounds. |
| Solicitors' fees (claimant) | Typically £5,000 to £30,000+ depending on track and complexity. Often exceeds the value of the claim. | None required. Parties use the platform directly. Lawyers may be instructed but are not necessary. |
| Solicitors' fees (defendant) | Comparable to claimant costs. Institutional defendants routinely spend more defending a claim than the claim is worth. | None required. The respondent uses the same platform. |
| Human decision-maker costs | Funded by the taxpayer. No direct charge, but reflected in court fees. | Reviewer fees set by the reviewer, from the low hundreds to £10,000+ per day for senior appointments. Parties choose their reviewer. |
| Adverse costs risk | Losing party may be ordered to pay the winner's costs. On the fast and multi-track, this can be tens of thousands of pounds. | No adverse costs save for the human reviewer's fees. No solicitors' costs to recover. |
| Enforceability | Court judgment. Enforceable by writ or warrant. | Arbitral award under s.66 Arbitration Act 1996. Enforceable as a court judgment. Recognised internationally under the New York Convention. |
| Process | Claim form, defence, directions, disclosure, witness statements, trial. Multiple interlocutory hearings possible. Adjournments common. | Twelve-phase online process. No hearings. No adjournments. No listing delays. |
In the US, the American Arbitration Association launched an AI arbitrator in November 2025 for construction disputes. Arbitrus.ai launched that year as an AI-first arbitration platform for B2B disputes. Bot Mediation launched an AI mediation tool. None of them operates under English law. None has a named panel of English lawyers and judges.
Those US launches confirm the market is forming. The gap in the UK is clear. We are building it now, before it fills.
ResolveThis runs disputes through a structured twelve-phase process, from the initiator's first contact to an enforceable arbitral award. Every step is online. No hearings. No court listings. No delays.
At its core, this is a paper-based arbitration. The platform automates the procedural steps that ordinarily consume months and tens of thousands of pounds in a conventional case: intake, evidence exchange, case management, and the production of a draft judgment. Evidence is gathered through structured AI interviews rather than at a trial, but the final product is the same as any arbitration conducted under the Arbitration Act 1996. The human reviewer receives a complete evidence bundle and a draft judgment, and produces a formal arbitral award.
There is no adversarial disclosure process. Each party relies on the documents it chooses to upload. A party may identify documents it believes the other side holds and ask for them to be produced, but there is no compulsion. If a party fails to produce a requested document, the AI may draw an adverse inference from that failure. This keeps the process fast and cheap, but it means the platform is not suited to disclosure-heavy disputes where the outcome turns on documents held by the other side.
The platform assesses case suitability at intake and will decline disputes that fall outside its current capability. At launch, the process follows a fixed set of procedural rules suited to cases of low to moderate complexity. As the platform develops, parties will have increasing flexibility to propose and agree procedures that suit them, including options for expert evidence, formal disclosure, multiple evidence rounds, and structured submissions.
| Phase | Action | Time |
|---|---|---|
| 1 | Initiator proposes use of the platform, selects procedural rules, and proposes a human reviewer | Initiator's pace |
| 2 | Other party agrees. Both parties pay platform fees. | 14 days |
| 3 | Complainant uploads documents, describes their complaint and the remedy sought | Complainant's pace |
| 4 | Respondent reviews the complainant's case, explains their defence, uploads documents, and states questions for the complainant | 14 days |
| 5 | AI creates list of issues in dispute | Automated |
| 6 | Complainant's structured AI interview, further documents, and witness evidence | Scheduled by platform |
| 7 | Respondent's structured AI interview, further documents, and witness evidence | Scheduled by platform |
| 8 | AI produces a single bundle of all transcripts and evidence | Automated |
| 9 to 10 | AI writes a reasoned judgment on the basis of the bundle | Automated |
| 11 | Losing party may appeal to human reviewer | On election |
| 12 | Human reviewer produces formal arbitral award under the Arbitration Act 1996 | On completion |
The initiator (who may be the complainant or the respondent) proposes the use of ResolveThis and selects the procedural rules from a set menu. At launch, this is a fixed format suited to straightforward disputes. As the product develops, the menu will expand to include more sophisticated options such as expert evidence, multiple evidence rounds, and structured disclosure.
The initiator also proposes a human reviewer from the platform's panel, which ranges from junior lawyers to KCs and retired judges, and agrees to be initially responsible for the reviewer's fees.
The other party then agrees to participate. When they click to accept, that click constitutes a binding arbitration agreement under section 6 of the Arbitration Act 1996. Both parties pay their platform fees at this point. Platform fees are calculated on usage: approximately 20p per page of evidence uploaded, 20p per minute of AI interview, and 50p per image or minute of video. This covers the token cost of the AI processing plus a 100% markup. Parties may use the platform directly or through their lawyers.
If the other party agrees and then disengages, the process continues in their absence and the resulting award is enforceable against them regardless.
The complainant uploads their documents and describes what they are complaining about and what remedy they want. The AI guides them through the process and flags anything that appears incomplete.
The respondent then reviews the complainant's case, explains their defence, uploads their own documents, and states any questions they want the complainant to answer during the interview phase.
On the basis of the complainant's case and the respondent's defence, the AI produces a list of the issues in dispute. This focuses the evidence phase that follows and ensures both sides know what is contested and what is common ground.
Each party undergoes a structured AI interview. This is the heart of the process. The AI probes inconsistencies in their account, asks follow-up questions arising from the documents, and puts the questions that the other side has requested. This replicates what cross-examination achieves in court proceedings, without the cost of a hearing. Each party can give evidence from home, by text or voice.
During their interview phase, each party may also upload further relevant documents and witness statements. Any witness may undergo their own AI interview, which is transcribed and added to the record.
The complainant suggests questions they want the respondent to answer. The respondent's interview follows the same format. The AI is trained to treat documents as genuine unless the other side expressly disputes them. Where a party alleges that a document is fabricated or a statement false, the platform prompts them to upload supporting evidence for that challenge.
The AI produces a single bundle containing all transcripts of interviews and all evidence that has been uploaded by both sides. On the basis of this bundle, the AI writes a reasoned judgment: background, issues for determination, findings of fact, applicable law, analysis, and outcome on each issue. The judgment is structured as a High Court judge would structure a reserved judgment.
The losing party may appeal to the human reviewer selected at Phase 1. The reviewer receives their agreed fee, the full evidence bundle, and the AI's reasoned judgment. They have full power to adopt, amend, or rewrite the judgment entirely. They then produce a formal arbitral award under the Arbitration Act 1996.
This is a legal document enforceable under section 66 of the Act and recognised internationally under the New York Convention. If the losing party does not pay, the winner can apply to court to enforce it as if it were a court judgment, without re-running the case.
Human reviewers are paid at agreed commercial rates, set by the reviewer and visible to the parties before the process begins. Even at the top end, the economics are compelling. A Deputy High Court judge charging £10,000 a day might spend two days reviewing a midsize case. That £20,000, plus the platform fees, represents the parties' total legal spend. No solicitors' fees. No counsel's fees. No interlocutory hearings. No disclosure costs. For a dispute that would otherwise generate six figures in conventional litigation costs, the saving is transformative.
It is a feature, not a bug, that the human reviewer will sometimes overturn the AI's judgment. This is a centaur model: human and machine working together. Appeal courts exist because judges miss things. So will AI. The human review layer is the quality guarantee, and the reason the outcome commands confidence.
We have produced a full demonstration determination. The fictional case involves a dispute between a homeowner and a building contractor over defective renovation works, covering breach of contract and the standard of workmanship implied by the Supply of Goods and Services Act 1982.
The determination runs to over 200 numbered paragraphs. It covers background, issues for determination, findings of fact, the legal framework, analysis of each issue, remedy, and an order. It is structured as a High Court judge would structure a reserved judgment.
The demonstration file is available on request. It is the strongest single illustration of what the platform produces and why it is credible as a substitute for court proceedings.
ResolveThis is designed for disputes where the facts are largely contained in the documents and accounts of the parties themselves. Tradesperson disputes, landlord and tenant claims, consumer complaints, unpaid invoices, freelance disagreements, service failures. These are cases where each side has its own documents and its own story, and what is needed is a fair, structured process to test both accounts and reach a decision.
The platform is not suited, at launch, to disputes that depend on extensive third-party disclosure, or where the outcome turns on documents held by the other side that cannot be obtained voluntarily. It is not suited to large-scale fraud, family law, or cases involving child welfare. The platform will decline these at intake.
In future, the platform will extend to more complex cases. Options for formal disclosure, expert evidence, multiple evidence rounds, and more sophisticated procedural rules will be rolled out as the product develops and as the AI's capabilities improve. The launch product is deliberately simple. That is a design choice, not a limitation.
The arbitral process is confidential. Neither party may disclose the proceedings or the outcome without the other's consent.
ResolveThis does not retain case data beyond the duration of the dispute. The platform is designed so that ResolveThis itself cannot access the substance of the parties' evidence. Data is processed for the purpose of the arbitration only and is not used to train or improve the underlying AI model. Parties may opt in to anonymised data retention at a discount, but this is never the default.
A "moat" is the term investors use for what makes a business defensible against competition. The technology behind ResolveThis is commercially available. A competitor could build a similar platform. What a competitor cannot replicate is the people, and what takes years to build is the commercial relationships.
The first moat is the training panel. The first priority is to recruit 20 to 30 named KCs, barristers, senior solicitors, and retired judges. These are ranked legal professionals, many of whom will go on to sit as judges. They will be responsible for training the AI in judicial reasoning and in the structured approach to weighing evidence that underpins every good judgment. That training is what gives the model its credibility. Their names and professional reputations will appear on the platform. A competitor cannot replicate that overnight.
The second moat is the reviewer network. ResolveThis will build commercial relationships with human reviewers who become familiar with the platform's process and are willing to offer fixed fees for a day or two's work: reading the evidence bundle, reading the AI's draft judgment, and producing what they consider to be a fair arbitral award. Sometimes that will be quick, because the AI's judgment is sound and the reviewer simply adopts it. Sometimes the reviewer will disagree with the AI and rewrite the judgment. That is the point of the system, not a flaw in it.
It will be for ResolveThis to manage those relationships commercially. Some reviewers will find the work predictable and rewarding. Others may find the AI's drafts require more correction than anticipated. As the AI improves, and the quality of its reasoning gets better, the reviewer's task will become lighter and their costs will decrease. The AI reasoning that large language models can produce today is already impressive. In five years, it will be better than most humans could manage. As reviewer costs fall over time, the platform's own margin grows.
A mobile phone company receives 500 claims a year from customers. The claims are for loss of internet access, consequential losses, and various grievances. Each claim is individually small, but they take years to pass through the county court system. The company faces a choice: pay panel solicitors significant sums to defend each one, or settle on a commercial basis, paying more than the claims are worth because the cost of fighting them is worse.
The company signs up with ResolveThis under a framework agreement. In cases it selects (those where it believes a claim is likely to be issued), it offers the would-be complainant the option of resolving the dispute through ResolveThis instead of court proceedings.
The complainant is happy. Their claim is dealt with within weeks, not years. They incur no legal costs. There is no risk of an adverse costs order against them, save in relation to the human reviewer's fees. And the mobile phone company is happy. It has eliminated panel solicitor fees entirely. The cost of the human reviewer is assigned to the losing party, and even that is in the low thousands of pounds due to the reviewer tier selected under the framework.
The same model applies to airlines defending luggage and delay claims, banks defending mis-selling complaints, and any organisation providing mass services to the public that generates a high volume of low-to-mid value claims each year.
The highest-value clients for ResolveThis are organisations that face a large volume of claims each year and currently spend disproportionate sums defending them. The target list includes:
Each institutional client that signs up can tailor the platform to its needs. They can integrate ResolveThis into their existing complaints process, so that disputes which cannot be resolved internally are referred to the platform automatically. They select the procedural rules from the available menu and choose human reviewers from the panel (who remain independent from the client, with no conflicts of interest). The framework agreement sets out the pricing, the default procedural rules, and the reviewer tier. From that point, referrals flow without further negotiation.
The platform requires a small operations team from launch.
A registrar (one full-time employee) will manage the day-to-day running of the platform. The registrar approves or rejects applications for extensions of time, handles administrative queries from parties, monitors case progress, and intervenes where procedural issues arise. The registrar is assisted by AI but is a real person making real decisions. The same person works directly with the development agency to identify and resolve technical issues as they arise, handles refunds, and manages any platform hitches in real time. This is a salary cost from day one.
A sales function will be needed to sign up institutional and business clients under framework agreements. The initial targets are the organisations listed above: banks, telecoms companies, airlines, housing associations, mass homebuilders, local councils, and similar. Each of these clients represents a pipeline of hundreds or thousands of disputes per year. The sales team does not need to be large, but it needs to be credible and commercially capable.
No UK company presently offers AI dispute resolution with binding outcomes and human review under English law. The US platforms confirm the market is forming. None operates in this jurisdiction. None has the named panel we are building.
The platform launches in England and Wales, where awards are enforceable under the Arbitration Act 1996. But international expansion is not a distant second phase. It is the goal from the outset. The platform is being designed for it.
The common law is trusted worldwide. Awards issued under the Arbitration Act 1996 are recognised and enforceable in over 170 countries under the New York Convention. Private parties in jurisdictions with slow, expensive, or unreliable local courts will trust an English-law platform backed by named English lawyers and judges. A Kenyan business in a contract dispute can access an English KC's judgment for a fraction of what local proceedings would cost. A Dubai-based company can resolve a commercial disagreement under English law without setting foot in London.
The economics of this business improve with every jurisdiction it enters. The platform, the AI, and the procedural framework do not need to be rebuilt for each new market. The human reviewer panel expands, but the core technology scales without proportionate cost. Each new institutional client in each new jurisdiction adds recurring revenue at high margin. That is why this is a scale business, not a services business.
Both founders are ranked barristers with extensive experience in dispute resolution. They understand how litigation works, how judges reason, and how to recruit the legal panel that is the core of the business. They will be directly involved in training the AI model in judicial reasoning and structured decision-making.
The next step is to recruit the wider training panel: 20 to 30 named KCs, barristers, senior solicitors, and retired judges. The founders' contacts across the legal market span most of the core legal disciplines. No judges have yet been signed up. That recruitment begins once the platform has a proof of concept and early backing. But the founders' own involvement in training the model, from day one, is itself part of the moat.
ResolveThis is not a law firm and does not need to be. It does not provide legal advice. It does not conduct litigation or exercise rights of audience. Those are the activities reserved under the Legal Services Act 2007.
ResolveThis is a dispute resolution platform. Parties agree to its terms, submit to its process, and receive a determination that is binding by agreement, not by statutory authority. The platform connects parties to human reviewers in the same way a marketplace connects buyers to sellers.
The regulatory position is clean. It was designed that way.
The domain is live. The process is designed in full. The demonstration materials are complete. Discussions have opened with people in tech and communications.
Before incorporating and committing fully to build, we are seeking conversations with early-stage funders. The build itself is not technically complex: document management, an AI reasoning engine built on commercially available models, a recorded evidence module, and a review workflow. None of this is novel.
The first use of funding is lawyer recruitment. A credible named panel of 20 to 30 KCs, barristers, senior solicitors, and retired judges changes this from a document to a business. That is what the first round secures.
Target launch: 1 January 2027. The resolvethis.co.uk domain is live.