In a conversation with Arthur Rothrock on The Litigator's Path, Letts breaks down why he thinks the billable hour is finally obsolete, how he's using automation and large language models to identify massive volumes of strict liability claims from public data, what makes Claude better than ChatGPT for legal work, and why most law firms are spending money on AI before identifying what problem they're trying to solve. The through line: AI adoption shouldn't terrify litigators – it should force them to stop billing by the hour and start capturing value in ways that align their incentives with their clients' interests.
Why Matthew Letts Left Private Practice to Build Four Legal Tech Companies
Letts qualified as a solicitor within months and was immediately recognized in the Legal 500 for insolvency work. But he left Isidore Goldman to pursue what he describes as "a shot I would regret if I didn't take" – a big data insolvency platform that addresses a problem he couldn't solve while billing by the hour. The core insight: there are massive volumes of undervalued insolvency claims that should be getting brought by entities like HMRC (His Majesty's Revenue and Customs), but those entities have all the data and do nothing with it. Letts saw the opening and built technology to exploit it. "In their incompetence lies my opportunity," he says.
The business model hinges on strict liability claims where all the data needed to build the case – and all the data needed to filter out statutory defenses – is available from public record. Letts and his Czech Republic–based technology team at Felix have pulled in over a billion data points from Companies House, the UK's corporate registry, and built what Letts calls "a giant magnet" that hovers over the haystack and attracts the needles. The system runs about 35 deterministic steps and one probabilistic step (a statistical analysis of how closely two data points match, which uses machine learning). The end result: Letts can identify claims at scale and pursue them himself rather than selling the data to someone else. "Once a litigator, always a litigator," he says.
Alongside the insolvency platform, Letts launched Codified Strategy to fill what he saw as a gap in the legal tech consultancy space. The problem: technologists with no legal experience are selling AI tools to lawyers who self-identify as Luddites, and no one is bridging the two sides with on-the-ground expertise. Letts positions himself as a "legal quant" – someone who brings the scientific rigor of his chemistry background to the intersection of law and technology. His consultancy embeds domain experts inside law firms to audit what's broken, identify low-hanging fruit in automation (not just AI), and build roadmaps that actually get implemented. The target market is SME law firms that treat IT like office equipment – a printer, a stapler – rather than a core strategic pillar. Letts has now hired his first full-time employee to expand the offering.
What Matthew Letts Recommends Instead of Harvey or Legora
When asked what tools firms should use if they're skeptical of big-name platforms like Harvey or Legora, Letts doesn't mince words. He describes those platforms as "99% marketing and 1% value" and notes that Harvey recently hired the actor who plays Harvey on Suits, which "tells you everything you need to know." His view: Harvey has been useful mostly as a way for large firms to dip a toe into AI in a safe, handheld environment staffed by former Magic Circle lawyers who can soothe their clients. But for firms that want to solve actual problems, Letts recommends starting with what the problem is – not with what the trendy solution is.
His first recommendation for AI-curious lawyers: start conversations with Claude, Anthropic's large language model. Letts calls the Opus 4.6 version "exceptionally good" for turning half-formed ideas into structured plans. He prefers Claude over ChatGPT because it's less sycophantic – it's less likely to agree with everything you say just to please you. When he wants to stress-test a litigation strategy, Letts gives Claude a persona: "Imagine you're the supervising partner acting for the other side. Tear this argument apart, find every weakness." The tool aims to please, so it aims to be that opposing partner and looks for holes in the strategy. If none of the holes it finds are real weaknesses, Letts knows he's on solid ground.
Letts also recommends that firms look inward before they look outward. What processes are highly repetitive? Which parts of the workflow are deterministic (you can write a flow chart with no ambiguity) versus probabilistic (there are judgment calls)? For deterministic problems, Letts says firms would be "somewhat mad" to use a large language model, which is a probabilistic tool. For probabilistic problems, firms might consider building their own small language model or retrieval-augmented generation (RAG) database so they can query their own case history and leading authorities without risking hallucinations from a general-purpose model. "It's just a really elegant rebadging of negligence," Letts says of hallucinations. His advice to senior lawyers: treat AI like an incredibly bright, keen, day-one trainee. You wouldn't sign on the bottom of their research and send it out the door without checking it first. The same applies to large language models.
How Matthew Letts Thinks the Billable Hour Will Finally Die
When Rothrock asks whether Letts is persuaded by the argument that AI tools will reduce billable hours and therefore reduce revenue, Letts is blunt: "AI should signal the death of the billable hour rather than the death of the lawyer using the billable hour." He calls the billable hour a "ready reckoner" for value exchange that has been overdue for replacement for decades. With AI reducing the amount of time required to produce work, Letts argues that litigators now have the opportunity to move to value-based pricing models like damages-based agreements (DBAs), the UK equivalent of contingency fees.
Under a DBA, a lawyer receives a fixed percentage of the damages the claimant recovers – 15% or 20%, for example – regardless of how much time the case took. The court doesn't care how many hours the lawyer spent. The client knows exactly where they stand: if we win, I give my lawyer 20%. Letts says DBAs are massively underutilized in the UK because firms see them as high-risk and worry about carrying work in progress. But if a firm can use AI to detach the amount it receives from the amount of time it spends, the risk profile changes. The firm and the client are clearly aligned, and the lawyer can spend less time on repetitive tasks and more time on high-value strategy.
Letts acknowledges that certain areas will remain sticky – defendant-side litigation, for example, where costs are still assessed on time spent. But he doesn't think lawyers have to stay wedded to the billable hour. "I'm not sure that any client ever asked for the billable hour," he says. "I think it just raises structural conflicts between incentives. Your lawyer's incentivized to spend more time to do the work." His view: if the billable hour survives, it will be through pure stubbornness of the profession, not because it's the best model.
Why Matthew Letts Says Most Law Firms Are Solving the Wrong Problem with AI
One of Letts's sharpest critiques is that many law firms – particularly larger ones – are spending money on AI because they have budget to spend, not because they've identified what problem they're trying to solve. "Lots of firms are essentially saying, well, AI is the solution before identifying what the problem is," he says. The result: firms buy expensive platforms that address a narrow part of practice and miss the low-hanging fruit in automation that's been available for 15 to 20 years.
Letts's consultancy starts by auditing the firm's processes. Where are work volumes concentrated? Which tasks are highly repetitive? Which processes can be written as a deterministic flow chart versus which ones require judgment? For deterministic tasks – things like document assembly, intake forms, or compliance checklists – Letts recommends automation tools that don't require AI at all. For probabilistic tasks, he recommends building custom solutions using RAG databases or small language models trained on the firm's own data. The key is to match the tool to the problem, not the other way around.
Letts also points to the broader ecosystem of legal technology that plugs into practice management platforms like Clio. Rather than buying a single monolithic AI platform like Harvey or Legora, firms can bolt on specialized tools that integrate with their core system and address specific pain points. The price point matters, too. Letts notes that a firm can get full-service functionality for less per seat per month than what Harvey or Legora charge, and the narrow scope of those platforms means firms are paying a premium for tools that don't cover most of their workflow.
The bigger opportunity, from Letts's perspective, is that firms can use automation to detach their ability to scale from their physical headcount. They can do more innovative work, be more profitable, and reduce the repetitive tasks that contribute to high rates of burnout in the profession. "Actually, there's so much opportunity to work collaboratively and to ultimately aim for that eureka moment where all the pieces come together," Letts says of the work he finds most engaging. Automation and AI should free lawyers to do more of that work – not replace them.
The Matthew Letts Crowdfunded Legal Challenge Against the UK Solicitors Regulation Authority
Letts made national news in the UK when he crowdfunded a legal challenge against the Solicitors Regulation Authority (SRA) over their treatment of a solicitor who survived a mental health crisis. The case involves a solicitor who had a difficult time at a law firm, reached out to the firm for help, and was fired. The solicitor reported the firm's conduct to the SRA, which has done nothing about the report to date. The firm then filed its own report about the solicitor, raising issues that had never been raised before. The SRA investigated and found no evidence to support the allegations, but during the process, the solicitor made an attempt on their own life. They survived.
The SRA obtained the solicitor's medical records and determined – without obtaining its own medical evidence – that the suicide attempt meant the solicitor posed a risk to the public and the profession. The regulator imposed conditions on the solicitor's ability to practice, prohibiting them from undertaking reserved or unreserved legal activities and barring them from working in any legal capacity in a regulated or unregulated setting. Letts describes the conditions as "astoundingly like a strike-off" – a removal from the roll – which the SRA doesn't have the power to impose.
The solicitor posted about their situation on Reddit. The post got significant engagement. Letts shared it on LinkedIn and started a crowdfunder. The case has now attracted pro bono support from junior counsel, King's Counsel, and a well-regarded international law firm. Letts says the solicitor is now ready to mount a challenge against the SRA's decision. The question is whether the solicitor would accept a settlement if the SRA reverses course or whether they would push for a formal decision to create precedent for others in similar situations.
Letts is passionate about the issue because he believes mental health shouldn't be weaponized against lawyers. "The alternative is you force people not to disclose that," he says. "And by the time that you find out someone's got a problem and then is struggling, they're at full mental breakdown stage." He argues that the legal profession has spent years addressing alcohol and drug abuse but is only now beginning to reckon with the systemic stress issues that cause those problems in the first place. "We're not actually addressing the root cause of all of this," he says. His view: creating safe, sustainable environments where lawyers can talk openly about mental health struggles will prevent far worse outcomes down the line.
What Litigators Can Take Away
-
Letts qualifies as a solicitor through the SQE route – two five-and-a-half-hour multiple-choice exams covering 16 areas of law – with no formal legal education, relying instead on a chemistry degree, scientific method, and a brutalist study approach that focused on passing, not perfection.
-
Letts uses Claude, not ChatGPT, for litigation strategy work and stress-tests his arguments by giving the model a persona: "Imagine you're the supervising partner acting for the other side. Tear this argument apart, find every weakness." If none of the holes it finds are real weaknesses, he knows his strategy is sound.
-
Letts argues that AI should kill the billable hour, not the lawyer – and that damages-based agreements (contingency fees) finally make economic sense now that firms can detach the amount they receive from the amount of time they spend on a case.
-
Letts's insolvency platform identifies over £1 billion in undervalued claims by pulling in over a billion public data points from the UK corporate registry and running 35 deterministic steps and one probabilistic step to filter out defenses and surface strict liability claims he can pursue himself.
-
Letts says most law firms are buying AI tools before identifying what problem they're trying to solve – and that the real opportunity is in automation that's been available for 15 to 20 years but hasn't been implemented because firms treat IT like office equipment instead of a core strategic pillar.
-
Letts crowdfunded a legal challenge against the UK Solicitors Regulation Authority after the regulator imposed conditions on a solicitor who survived a suicide attempt, effectively barring them from practicing law without the power to formally strike them off – a case he sees as part of a broader fight to ensure mental health isn't weaponized against lawyers.
Listen to the Full Episode
Hear the full conversation with Matthew Letts on The Litigator's Path at legion.law/podcasts or search for "The Litigator's Path" wherever you get your podcasts. For more on Legion's litigation document drafting platform, visit legion.law.