Lessons from 40 Years of Building Agents

TED and the Future of the Lawyer

I meet Nishat Ruiter, GC at TED, the world's most influential ideas platform, and learn how TED Law is asking us to question what it means to be a lawyer. What we cover * How TED grew from a conference to become the world’s #1 ideas platform. * Nishat’s unconventional path, via the Netherlands, the Brooklyn family court, and the self-taught IP licensing work that helped launch her GC career. * How organising a local TEDx event gave Nishat a deep understanding of TED’s mission before she ever joined the organisation. * What TED looks like behind the scenes, from the main conference to TEDx, TED-Ed, podcasts, partnerships, licensing, translation and global community-building. * Why Nishat launched TED Law, and what she thinks legal education misses around judgment, professional identity, cultural competence, collaboration and moral courage. * How you can get involved in TED Law (that includes law firms, in-house teams and tech companies). Meet Nishat When I catch up with Nishat Ruiter, it has only been a few weeks since the latest TED conference in Vancouver. The conference itself is over, which sounds like a natural moment to come up for air, but TED is really a year-round organisation of talks, TEDx communities, podcasts, partnerships, education programmes, licences, brand questions, rights questions, publication decisions and all the legal work that has to happen to make sure ideas can safely reach the world and have the maximum impact. Nishat sees that world from an unusual seat. She gets the force of the talks, the audience energy, the spectacle and the sense of possibility, while also leading the back-office legal review that makes the public version possible. A TED Talk may look effortless by the time it reaches the stage or the internet, but behind it sit questions about rights, privacy, defamation, intellectual property, accuracy, context and reputation. That is part of what makes her role so interesting: she is protecting ideas without wanting to smother them. The bigger question in our conversation is what happens when you apply TED’s vision of “Ideas change everything” to the world of law. TED is built around curiosity, clarity, public communication and the belief that a well-framed idea can move people. Law, at its best, is built around judgment, service, justice and the structures that allow human beings to cooperate. Yet the profession has often taught lawyers the rules far more deliberately than it has taught them how to carry those rules into leadership, crisis, culture, technology and moral choice. TED Law is Nishat’s attempt to address that gap. From Dracula to Brooklyn family court Nishat did not grow up with a plan to become a lawyer. Her first love was theatre. In high school, she acted seriously enough to play the lead in Dracula, which I point out should definitely be on her LinkedIn profile. She was drawn to the empathy of acting: understanding a character, getting under the surface, working out why someone acts as they do. A high school director then gave her a blunt assessment of the acting world. If she wanted to pursue it, she should understand the reality of the industry: she was not white, she was not blonde, and she would not fit easily into the box. It was a hard thing to hear, and it became one of the early experiences behind a theme that runs through her career: finding herself outside the expected shape and then building from there. Law arrived almost sideways. In college, she took a course on the legal and social environment of business. The professor posed a simple hypothetical: a student slips on a banana peel at a university. Who is responsible? Everyone wanted to defend the student. Nishat raised her hand to defend the university, less because she had some sophisticated view of liability and more because she was excited by the less obvious side of the argument. She liked the reasoning, the exploration, the challenge of working out the answer. That instinct took her to law school, where she initially imagined a career in international human rights and justice. After graduating, she spent time in the Netherlands, where her husband is from, and explored international work. But law school loans and practical realities eventually pulled her back to the United States. When she returned, she did what many young lawyers do: she applied for what was open. One of those openings was in family law. She worked for a sole practitioner in Brooklyn for two and a half years and saw, up close, the human pressure of matrimonial litigation. Custody. Support. Divorce. Lives being rearranged in courtrooms that had far too many cases and far too little time. At one point, she says, a judge might have 120 cases on the docket and two minutes to hear yours. The work mattered enormously, but the system around it could feel too arbitrary for questions that were so consequential. So she moved back toward intellectual property and technology. This was the dot-com era, before anyone could ask ChatGPT how to become a software contracts lawyer. Nishat went to Barnes & Noble, read everything she could find, and built herself a three-ring binder full of software licensing contracts. When she interviewed for a contracts manager role and they said they were looking for someone with templates, she had them. Then she made the obvious legal career move: if she was going to be the only legal person, perhaps she should be general counsel. That is a very Nishat story. She sees the gap, does the work, builds the thing and then makes it happen. Thanks for reading Agents.law! Subscribe for free to receive new posts and support my work. Finding TED before TED found her Before TED, Nishat built a career across IP, fintech, software, legal tech and a decade as associate general counsel at CA Technologies. The bridge into TED, though, began in Hillsborough, New Jersey, where she and her husband had chosen to live for practical reasons: schools, work, family life. They also cared deeply about community. Her husband is Dutch. Nishat is Pakistani. Between Dutch bluntness and Pakistani directness, she says, there was no shortage of intense discussion at home, but there was also a shared global outlook and an interest in ideas that could bring people together. Nishat was serving on the township library board and felt that the community could do more to surface meaningful ideas. TED had already been part of their family life. Her husband was an early fan, and they would introduce high school students to TED Talks. So they organised a TEDx event through the library. That meant finding speakers, shaping talks, dealing with the intellectual property, building the event and learning what TED actually requires of a local community. It took six months. She did it without any thought that she would one day work for TED, but it gave her a lived understanding of the brand, the mission and what happens when TED’s global idea is translated into a local setting. Then, during a career shift, the pieces came together. Nishat had just accepted a new role in New Jersey. Her children were in high school. She was tired of commuting into New York and wanted to be more present at home. Around the same time, a friend who had recently joined TED in video and TED Talks asked about her work. At the time, TED had no in-house lawyer. Outside firms helped with some matters, but the publication of talks raised exactly the kinds of questions Nishat understood: IP licensing, defamation, privacy, rights, risk and the judgment needed to review content without flattening the idea behind it. She had already accepted another job, so when she sent over her resume she added the caveats: she had family constraints, she did not want a five-day-a-week New York commute, and the timing would be difficult. Chris Anderson wrote back the next day. Within a week, she had met with TED’s leadership and had an offer. It sounds almost too neat, but the fit makes sense. TED needed a lawyer who understood intellectual property and also understood why the ideas mattered. The platform behind the talks Most people know TED through the talks. For many of us, TED is still associated with the canonical talks that seemed to break through the early internet: Sir Ken Robinson on education and creativity, Hans Rosling on population and data, and countless others that made complicated ideas feel alive. But TED is much broader than the stage. It began in 1984 as a conference around technology, entertainment and design. Chris Anderson later helped turn it into a nonprofit and, in 2006, TED began putting talks online. The expectation was modest. Perhaps the talks would reach 100,000 views. They reached around a million in the first month, and TED became a way of bringing ideas to people wherever they were. That required production, publishing, translation, distribution, licensing, brand stewardship and, eventually, an enormous global community. TEDx was another leap. What began as a risky experiment in giving communities a free licence to convene around ideas has become something like 4,000 to 4,500 events around the world. Then there is TED-Ed, with its animated lessons created with teachers and artists. There are podcasts, partnerships, education programmes, the Audacious Project and the many other ways TED tries to make ideas travel. A small legal team supports all of that. Their job is to help TED stay true to itself while operating globally, across formats, partners, communities, rights systems and cultural contexts. It is harder than it looks, and it gives Nishat a useful vantage point on the difference between protecting an institution and preserving the spirit that made the institution worth protecting in the first place. AI and the purpose question We spoke, inevitably, about AI. TED has always been close to the frontier of technology; many of the people shaping what comes next have appeared on the TED stage early, sometimes before their ideas became mainstream. The most compelling use of AI, in her view, is the one that serves humanity: improving education, science, health, access and the ability of human beings to do more meaningful work. She points to global discussions about AI for humanity as the kind of framing that deserves more attention. That becomes the bridge to TED Law, because lawyers are facing the same problem in their own field. A technology that can change the mechanics of work forces a deeper question about the purpose of the work. If AI can draft, summarise, search, structure and increasingly act, the lawyer’s value has to move beyond information retrieval and formal rule application into judgment, courage, context, communication and the ability to make sense of human stakes. Why TED Law exists TED Law grew out of several pressures arriving at once. Some were legal and political. Nishat points to the overturning of major precedent in the United States, including Roe v. Wade, and the broader conversations that followed about rule of law, institutions and what lawyers are meant to do when foundational assumptions are tested. Some were professional. In conversations with general counsel, partners and legal leaders, she kept hearing the same unease: the skills lawyers were using in leadership were often different from the skills they had been taught in law school. Some of it was personal. Nishat remembers staying after class in law school because she wanted to ask why. She wanted to understand why a judge had decided a case in a particular way, what was happening in the community, what the broader context was, and what was going on in the world around the decision. Another strong student once stayed behind too, then asked the professor whether any of that context would be tested. The answer was no. It was just context. The student left. For Nishat, that moment captured something important about legal education. The profession teaches lawyers to read, extract, repeat and advocate. It is less consistent in teaching them to connect law to values, identity, culture, judgment, leadership and courage. That may be manageable in parts of legal practice, but it becomes a serious gap once lawyers move into leadership roles and face moral questions, crisis decisions, compliance failures, rule-breaking, political pressure and business demands. At that point, lawyers sometimes have to push back on the very people they advise. They have to represent an organisation while retaining their own professional judgment. Substantive law still matters, obviously. But Nishat’s point is that it no longer carries the whole burden. TED Law is built around capabilities the profession needs to teach more deliberately: professional identity, critical thinking, cultural competence, intuitive collaboration and the ability to work effectively with AI. The result is a curriculum of roughly 15 to 18 hours, built to pair TED’s ability to open up a subject with the legal rigour needed to examine it properly. The work has been tested against the profession rather than designed from a private theory of what lawyers need. TED Law surveyed law schools, general counsel, managing partners, law students and lawyers across countries. It set up advisory councils. It looked at regulatory changes, including ABA developments around professional identity, client interaction and cultural competence. The gaps Nishat had identified were already showing up in the profession’s own requirements. Teaching lawyers without another terrible slide deck Anyone who has sat through enough legal training will understand part of the problem. Lawyers are cynical and sceptical audiences, and many CLE-style sessions invite exactly that response: dense slides, tiny font, rules read from the screen, and a few seconds at the end for questions so the box can be ticked. You leave with the materials and the credit, but rarely with a changed view of the work. TED’s method gives Nishat a different tool. A talk can change the room. It can lower the guard, create emotional entry and move lawyers from defensiveness into curiosity. From there, you can have a serious legal discussion: a case study, a conversation about ethics, AI, culture, judgment, practice readiness or professional identity. The inspiration is not the whole lesson; it is the opening that lets the lesson breathe. That may sound soft to some lawyers, but the wellbeing data tells a harder story. Nishat is direct about the fact that lawyers experience suicide ideation at around four times the rate of other professions, alongside higher levels of alcohol abuse and drug abuse. Research also suggests that the things that make lawyers feel fulfilled are autonomy, relationships and meaning rather than prestige, rank or power. The legal profession has spent a long time optimising for markers that do not necessarily make lawyers better, healthier or more useful. TED Law asks a different set of questions: why did you come to law in the first place? What did you write in your law school application essay? Was it really about money, status and golden handcuffs, or was there once an idea about justice, service, purpose and doing something useful with a powerful set of skills? The AI-age lawyer All of this has become more urgent because of AI. ChatGPT became publicly available in late 2022, and in the short period since then the legal industry has moved from basic experimentation to agentic AI, automation, workflow redesign and increasingly serious questions about what legal jobs will look like. None of us knows exactly what the profession will look like in two or three years. That uncertainty changes the education question. If the rules, tools and job categories of the near future are still forming, the foundations have to carry more weight: critical thinking, collaboration, creativity, cultural competence, multidisciplinary judgment, moral imagination and the ability to use AI without surrendering human responsibility to it. Nishat is interested in something deeper than skills training. She wants lawyers to examine their role in society and reach for a better version of the profession’s potential. If lawyers do not understand human beings, she asks, how are they supposed to help build the structures that protect them? An incubator with serious ambition TED Law is still in incubator stage, which is an important distinction. The work is being built, tested, refined and rolled out carefully. People can sign up for updates through TED’s legal newsletter at ted.com/law, and Nishat is actively inviting input through surveys and community conversations. There are already concrete paths in motion. TED Law is working with the ACC Foundation, the nonprofit arm of the Association of Corporate Counsel, to bring training to in-house lawyers. It is also working with the American Bar Association, including an in-person training in Chicago in July. Nishat wants lawyers everywhere to understand their professional values and identity, think critically, collaborate intuitively, operate across cultures and use AI effectively. Whether TED Law becomes the vehicle for all of that is secondary to the impact she wants to see. She also wants to change the way society sees lawyers. The old Shakespeare line about killing all the lawyers is usually treated as a joke at the profession’s expense. Nishat points out it actually had a different meaning: remove lawyers and you remove a mechanism for addressing injustice. That is the role she wants lawyers to recover. Final note What struck me most about Nishat is that TED Law feels like the convergence of the whole story. The theatre student who learned to inhabit a character; the young woman told she did not fit the box; the law student who wanted to know why; the Brooklyn family lawyer who saw the human cost of an overstretched system; the IP lawyer who built herself a binder because nobody was going to hand her the answer; the community organiser who brought TEDx to her town; the general counsel who understands that ideas need support. The project is still young, and Nishat would be the first to say it is being built with others: four advisory councils, law schools, GCs, law firms, students, professional bodies and lawyers who know something is missing even if they have not always had the language for it. But the central idea is already clear. The future of law will require lawyers who know what they stand for, understand the people they serve and can build with technology without forgetting why the work matters. That is a big idea, and TED is a fitting home for it. Resources * Sign up for the TED Law newsletter [https://www.ted.com/law] * Sir Ken Robinson and the best TED Talk ever (in my opinion!) [https://www.ted.com/talks/sir_ken_robinson_do_schools_kill_creativity?hsa_acc=7777130675&hsa_cam=218945593&hsa_grp=17379703273&hsa_ad=295607643496&hsa_src=g&hsa_tgt=kwd-16365900407&hsa_kw=best+ted+talks&hsa_mt=b&hsa_net=adwords&hsa_ver=3&gad_source=1&gad_campaignid=218945593] * Follow Nishat on LinkedIn [https://www.linkedin.com/in/nishat-ruiter-1a21362] This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.agents.law [https://www.agents.law?utm_medium=podcast&utm_campaign=CTA_1]

The World’s Best Vibecoder is a Lawyer

In the latest episode of Without Limitation, I meet Michael T. Brown. When I sit down with Mike, it has been around three months since he won the Anthropic global hackathon, beating 13,000 applicants, most of them career engineers, with a project built in six days from his home in Orange County. We’re also chatting on the day before he takes the main stage at Code with Claude in San Francisco, one of the biggest software events in the world this year, where he will be delivering a thirty-minute keynote. Listen to the full episode on your favourite platform, or keep reading for my take on the conversation. * Apple Podcasts [https://podcasts.apple.com/us/podcast/without-limitation/id1870080229] * Spotify [https://open.spotify.com/show/3JP46jTirPmEan6TfuVW7k] Not a software engineer Mike is, by his own admission, not a software engineer. He has never written a line of code in anger. He is a personal injury attorney by background, with a previous career in Hollywood. Of the top 5 finalists that Anthropic announced, only one was a full-time software engineer. The others included a cardiologist, an electronic musician, an infrastructure worker from Uganda, and Mike, who took the overall crown. So how did a lawyer win the Super Bowl of hackathons? And what does his story tell the rest of us about what is possible with these tools right now? Join me on 17-18 June: We are running what promises to be the biggest vibecode hackathon in law, at LegalTechTalk in London. This is a unique event, in partnership with HSF Kramer and Replit. You’ll get free Replit Pro credits just for participating, and there are some great prizes on offer. But to be involved, you must register [https://www.legaltech-talk.com/experience/legaltechtalk-vibeathon/]. Hollywood to law school to a maxed-out credit card Mike didn’t set out to be a coder, or a lawyer for that matter. He spent his twenties in Hollywood doing visual effects work. He did what he describes, with a smile, as the rational response to any burnout: he went to law school. He fell into personal injury practice and genuinely liked it. He spent a few years at California PI firms before hitting the decision point most PI lawyers eventually hit, which is making the decision to go out on his own. That decision happened to land in late 2023, just as ChatGPT was changing what was possible. Faced with the usual choice between hiring paralegals or burning through savings on back office support, Mike did something different. He started building prompt libraries before anyone was calling them prompt libraries, recording intake calls with client permission, transcribing them through Whisper, and feeding the transcripts into ChatGPT to produce his own version of an intake form. The system worked well enough that his cousin, a serial entrepreneur, suggested they productise it. They built a company called Onbreeze, a note-taking phone line for lawyers, on the insight that injured clients don’t book a Zoom call with a lawyer - they call. Vibecoding wasn’t mainstream at this point, and Mike outsourced the entire development process to an external development agency. The product launched in January 2025. Unfortunately for Mike, it crashed the moment a fourth caller hit the line. When the offshore engineering team explained that fixing it would cost roughly the same as building it had, Mike was already maxed out on credit cards. He downloaded Cursor (an AI coding tool) instead. Ten minutes later, by talking to it about what he needed, he had a working solution. He still couldn’t read the code. But the thing worked. That was his first prompt. How Mike learned to build Onbreeze didn’t take off the way Mike hoped. By May 2025 he was at a crossroads: go back to practising law full time, or lean into the strange new fluency he’d developed with vibecoding tools. He chose the tools. But what’s interesting is how he chose them. He didn’t go and learn to code in the traditional sense. He took on a series of consulting projects for friends and family, each of which forced him to learn one new thing. A roll-up project for a private equity contact taught him how to handle data across complex databases. A pickup ordering system for his cousin’s butcher shop in Brooklyn taught him Stripe. Mako, a demand letter generator for personal injury lawyers, taught him how to deploy agents to the cloud using Anthropic’s SDK. He describes it as accidentally building a curriculum. Each project was something he wanted to learn anyway. Each one came with a real user on the other side. And each one added to a stack of skills he could combine later. By the time the Anthropic hackathon was announced in early 2026, Mike was already far more fluent with the technology than most. Picking a problem nobody could solve The hackathon accepted only 500 of the 13,000 applicants, and the brief was specific: they wanted problems that weren’t solvable with previous generation models. The problem was California’s housing crisis, or more precisely, the permits bottleneck inside it. “Everyone thinks California has a housing crisis,” Mike puts it. “We don’t. We have a permit crisis.” His friend builds ADUs (Accessory Dwelling Units), the small backyard cottages California has incentivised as a partial answer to its housing shortage. The laws are favourable, the demand is there, and yet permits routinely take six to nine months, with endless back and forth between builders and city plan reviewers. Mike had a hunch AI could help. He asked Cameron to send him a set of blueprints. Claude couldn’t handle them. The pages were physically too big, the size of a table, and the dense annotations meant traditional OCR pulled the text away from the visual context that gave it meaning. A note about wall thickness only mattered if you knew which wall it referred to. By this point, most people would probably move on, but Mike saw an opportunity. If the most advanced Anthropic model available at the time couldn’t handle this, then it was exactly the kind of problem the hackathon judges were looking for. He submitted, got accepted, and immediately had a small panic about whether he could actually solve it. The lesson buried in this is that problems are way more important than solutions. Most vibecoders jump straight into building things. Mike spent his time finding a problem worth building for, one that mattered to a real user and that pushed the technology to its current limit. Building it in six days Mike decided to call the app CrossBeam. The hackathon ran for a week and by Wednesday night, two days in, Mike was in trouble. He had two disparate approaches, neither of them working, and the hour was getting late. He scrapped the elaborate combination of OCR plus image referencing he’d been wrestling with, and went back to first principles: here are the rules, here are the chunks of blueprint, figure out what’s wrong. That first end-to-end run hit roughly 40% accuracy against the city’s correction letters. Crucially, it ran. Once a system runs end-to-end, even badly, you can benchmark it and improve it. By Thursday afternoon he was at 60 to 70%. Two design choices were key here. The first was treating the model’s context window as a precious resource rather than a dumping ground. Mike used Skills, Anthropic’s open standard (think of them like those files that get uploaded to Neo’s brain in the Matrix so he suddenly knows Kung Fu), to keep his prompts lean. He used parallel sub-agents, each with a narrow job and a fresh context, rather than one bloated agent trying to do everything. He used what he calls adversarial prompting: have one agent plan, a second agent execute, and a third agent test, so that no single agent is grading its own homework. For this, he uses an analogy from legal practice: “When an associate brings you a memo, you don’t ask them if they double checked it. Of course they’re going to say yes. You give it to someone else to check.” The second design choice was testing. He set up command line interfaces on the backend so that one Claude instance could run a new ADU through the system every hour while he worked on something else, tracking accuracy gains in real time. By Friday his accuracy was high enough that he could spend the weekend on the user-facing piece, which itself ended up being shaped by two conversations. The first was with Cameron, his friend, a builder. Mike sent him fresh outputs and Cameron immediately asked if he could run three more blueprints through. Real users behave like that when something works. The second was with his friend Connor Traut, the mayor of Buena Park. Until that conversation, Mike had only been thinking about the builder side. Connor flipped his perspective: cities also need help, on the review side, processing permits faster. So Mike rebuilt the product to serve both sides. Builders drag and drop blueprints and get back a precise action plan in twenty minutes. Cities batch process submissions and generate draft correction letters. Same product, two different audiences. Why lawyers might be unusually good at this Mike’s prompting structure is one most lawyers will recognise immediately. Issue, rule, analysis, conclusion. This is the framework you learn for the bar exam, repurposed as the framework for getting good output from a frontier model. Mike starts with the issue at the top. Then, the relevant rules and constraints. Next, the analysis: why this matters, who it’s for, what good looks like. Conclusion at the end. He often prompts by voice, often for five minutes at a stretch, because typing caps the average user at around 60 words a minute and speech runs closer to 180. (Ahem those of us who took a typing class in the 90s are closer to 120+ but the point stands!) The skillset Mike describes is closer to drafting a memo than to writing code. Lawyers spend years learning to think clearly under pressure, to anticipate counterarguments, to specify exactly what they want and exactly what they don’t. Those skills, plus a working “b******t detector” and a willingness to experiment, are most of what it takes to build with these tools today. The bit lawyers usually need to add is the last part - the curiosity and willingness to experiment. Some legal training does lean towards risk averseness in a way that doesn’t serve help with building new things. We may need to unlearn some things here. Thanks for reading Agents.law! If you’re enjoying this piece, please share it with someone else who may find it useful. On security As we talk, Will Chen’s MikeOSS project has been making waves. It’s an open source version of Harvey/Legora. Mike and I observe that the conversation on vibecoding has become a little polarised into those who seem to believe you no longer need to learn software engineering at all (on the one hand) and those who believe it is a waste of time (on the other). Mike is realistic here, if a little more bullish than I am on AI’s ability to spot and resolve security and compliance issues in production software at the current time. He’s alive to the limitations of vibecoding and shares that he would be open to partnering with software engineers when the project demands it. (Side note: my recommended framework for Responsible Vibecoding is at Vibecode.law/learn [https://www.vibcode.law/learn]). His biggest frustration is people writing about the issues with open source software rather than submitting PRs to resolve the issues. This ties into something else I observe about Mike - his extreme bias towards action. He doesn’t have time for essays about software - he wants to build. Key lessons for vibecoders If we step back from the specifics of permits and ADUs, I think what you end up with is a template for building effectively with AI. Find a problem worth solving. Talk to the people who actually have it and spend as much time with them as you can. The closer you can get to the problem, the better the output. No problem, no solution, no value. Build in short cycles with rigorous testing. Testing is a critical part of software engineering but it’s something that most vibecoders forget. Software can be tested really effectively because it largely either works or it doesn’t. Don’t forget this part. Treat the context window like a budget. Mike learned through experimentation how performance drops off a cliff as you use up the agent’s context window. He manages this carefully with compaction, Skills and just starting new conversations. Prompting still matters. I occasionally read that the era of prompt engineering is coming to an and. I don’t agree on this. Sure, some of the old hacks like “You are a world expert in everything” aren’t very useful these days but Mike’s framework for prompting is effectively about giving the agent the exact context it needs. This leads to better outputs and CrossBeam is proof. Show your work to real users early and let their reactions reshape the product. Don’t be precious about your first design. Most vibecoders spend so long seeking perfection and not enough time getting it in the hands of real users. Finally, be curious. This is the common theme across all episodes on the podcast. People like Mike who are pushing the boundaries of what is possible do so because they are curious about the boundaries. Most people thought the California permitting crisis couldn’t be solved with AI. Mike wondered what would happen if it could. I for one can’t wait to see what Mike builds next. Up next In the next episode, I’ll be sitting down with the amazing Nishat Ruiter, General Counsel of TED and the founder of TEDLaw. Resources & further reading Follow Mike on LinkedIn [https://www.linkedin.com/in/michael-t-brown-034aaa22/] CrossBeam [https://www.crossbeam-permits.com/] Thanks for reading Agents.law! Subscribe for free to receive new posts and support my work. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.agents.law [https://www.agents.law?utm_medium=podcast&utm_campaign=CTA_1]

Why India is a Legal Tech Superpower

Shreya Vajpei built the community for legal tech in India. Now she's connecting that community with the rest of the world. India has more legal tech startups than almost anywhere on earth. I did not know this until I sat down with Shreya Vajpei, but India has around a thousand legal tech startups, which Shreya tells me puts it second only to the US and ahead of every other market, including the UK. That is a striking number for a country where foreign law firms still can’t really practise, where only advocates can own law firms, and where the entire legal services market is about a fifth the size of the UK’s. If there’s one person who can help us make sense of this landscape, it’s Shreya Vajpei. Listen to the full episode on your favourite platform, or keep reading for the full write-up. * Apple Podcasts [https://podcasts.apple.com/us/podcast/without-limitation/id1870080229] * Spotify [https://open.spotify.com/show/3JP46jTirPmEan6TfuVW7k] Introducing Shreya Shreya trained at Khaitan & Co, one of India’s tier one firms and roughly the magic circle equivalent in the Indian market, with around 800 lawyers when she joined and closer to twice that today. Like many guests on Without Limitation, her career has taken an unconventional path. She practised for a couple of years before moving into a practice development role. Then, COVID hit and the marketing budget disappeared overnight, which meant her team ended up absorbing everything else the managing partner needed help with. That covered IT, HR and operations, but also pricing, strategy, new office openings, partner performance and strategic hiring. In her words, whatever landed at the managing partner’s table also landed at theirs, which gave her something most lawyers never really get, which is a top-down view of a law firm as a business rather than a bottom-up view of a practice. From there she became one of the first hires into Khaitan’s innovation team, and her role kept evolving as AI did. Last year she moved to the UK to join Stephenson Harwood, drawn by what she described as a more mature market for digital transformation, with longer-established innovation teams, more consistent IT budgets, and a decade or so head start on the journey. A thousand startups Shreya tells me that India’s legal services market is around $10 to $15 billion in total, roughly a fifth the size of the UK’s, and it is wildly unconsolidated compared to what most of us are used to. The top five firms are similar in size to each other, in the 1,500 to 2,000 lawyer range, and then there is a big gap before you reach the long tail of full-service firms operating somewhere between 50 and 200 lawyers, and then the boutiques and independent chambers, and then a hyperlocal market across the tier two and tier three cities that operates almost entirely separately and accounts for the vast majority of India by population if not by revenue. On top of that, foreign firms still can’t really practise there, a liberalisation bill has been pending for years with significant local opposition, and there is no ABS or non-lawyer ownership of any kind. So the obvious question is why a market with all of those constraints has produced so many startups, and Shreya’s answer is partly cultural, partly economic, and partly a story about talent. India, she tells me, is generally entrepreneurial and high risk-taking, which feeds directly into the volume of founders willing to have a go, and it has some of the best developers in the world at cost structures that make building viable in a way that is hard to replicate elsewhere. The economics also push founders outward almost from day one, because rupee revenues are small once converted, so the dominant playbook is to build in India and sell internationally, which is exactly what companies like SpotDraft (now established in the US and entering Europe) and Lucio (which recently opened a New York office) have done. One category worth flagging, because it is more advanced in India than most other places, is online dispute resolution (ODR). SEBI, the securities regulator, now requires all investor disputes to go through an ODR platform, and there is open API infrastructure called the Pulse Protocol that allows any ODR provider to plug in, in much the same way that UPI revolutionised payments by giving every bank and every app a shared rail to build on. India tends to solve problems at the infrastructure layer when it solves them properly, and ODR is a good example of what that looks like in practice. The bridge What makes Shreya interesting beyond her own story is that she has lived on both sides of the bridge between the Indian and international legal tech ecosystems, and she has clear views on what each side keeps getting wrong about the other. For Indian startups looking to scale into the UK and US, she offers a network and an instinct for what magic circle firms actually buy, which is not always what an Indian founder might assume from a distance. Last year she worked with the UK Department of Business and Trade to bring a contingent of Indian legal tech startups to the UK to meet magic circle firms, Scottish firms, and the Legal Tech Talk crowd, and that kind of bridging is something the industry could probably use more of. For international players trying to enter India, she sees the same mistakes repeated. Pricing set for the US market with no real adjustment for local realities. Customer support sitting in time zones that do not overlap with the Indian working day. A lack of appreciation for the fact that the biggest Indian law firms still operate across multiple languages alongside English, and that most LLMs do not handle Indian languages or scripts particularly well, which means translation is a first-order use case rather than an afterthought. She makes the point that the legal AI players doing well in India have generally understood that the same product positioning does not travel intact, because the problem itself is not quite the same as it is in the US, and the reframing has to happen locally rather than being assumed away. If you want to tap into this network, it is at indianlegaltech.net [http://indianlegaltech.net]. On influential women in legal tech Shreya recently won an ILTA Influential Women in Legal Tech award. We discuss how her award sits against a backdrop that most people in the industry recognise but that not enough are actively doing something about, which is that around 3% of startup funding flows to female founders, and legal tech is no exception. As Shreya puts it, some parts of the ecosystem have become a “boys’ club” where the funders and the people asking for funding are both part of the same cycle, and that cycle is genuinely hard to break from inside it. Her recommendations are practical rather than abstract. She suggests that if your firm runs an incubation programme, ring-fence dedicated seats for female founders. If you sit on a legal tech fund, write a hypothesis that requires a female founder on every backed team. And if you are a woman who has made it into a decision-making role with some capital to deploy, invest in another female founder, because even one extra month of runway can sometimes be the difference between a company that survives and one that doesn’t. She also made an observation that that a lot of the buyer-side decision makers in legal innovation, the heads of innovation and heads of knowledge across the larger firms, are women, and the supply side of the industry has not really caught up with that. On the AI-native firm We finished, as I tend to with these conversations, on whether law firms are actually changing or just dressing the old model up in new clothes. Shreya uses the factory electrification analogy, which is my personal favourite as well. When factories first switched from steam to electric power, owners swapped the engines but kept the same layout, the same processes and the same workflows, which meant they got slightly faster operations but not much else. The real productivity gains came decades later, when factories were redesigned from the ground up around the new power source rather than just retrofitting it into the old design. Most law firms are still firmly in the swap-the-engines phase, treating AI as a tool that makes existing tasks slightly more efficient rather than as a medium for rethinking what legal work is and where the value actually sits. Look at almost any legal tech company website and the framing is the same: draft 20% faster, research 30% faster, all of which assume the underlying tasks remain the tasks lawyers do. The harder question, which Shreya thinks almost no firm has properly sat down with, is where the value layer actually lives in a professional services business once AI is properly in the picture, and what you would build if you started from that question rather than from the current shape of the firm. AI is now forcing the rethink that should arguably have happened years ago, and for those of us interested in true innovation in legal services, that is probably the most exciting part. If you are building in legal tech and you have not yet thought seriously about India, or if you are in India and thinking about scaling out, Shreya is the person to know. I look forward to seeing what she builds next. Links * indianlegaltech.net [http://indianlegaltech.net] * Shreya on LinkedIn [https://www.linkedin.com/in/shreyavajpei] * Stephenson Harwood [https://www.stephensonharwood.com/about/innovation/] This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.agents.law [https://www.agents.law?utm_medium=podcast&utm_campaign=CTA_1]

Lessons from 40 Years of Building Agents

Dazza Greenwood has been building agents for longer than several legal tech founders have been alive. In the 1980s, as an undergraduate computer science student, he encountered AI assistants for the first time. His module introduced a then-new paradigm: human language, chat-based systems. The exercise was to build something modelled on ELIZA, the MIT chatbot whose therapist module ran on a simple heuristic. Find the keyword. Reflect it back to the user. When in doubt, tie it all back to the user’s parents. It was deterministic, a little absurd, and wildly popular with the people who tried it. Dazza learned some tricks and came away with a fascination with what AI was, and more importantly, what it could be. Note: Some of the concepts in this episode may be unfamiliar to some listeners. We cover them in the technical explainer at the end. What followed was a career which has spanned dozens of initiatives around the world. Let’s just say that Dazza has worn (and continues to wear) a lot of hats. Legislative aide. Candidate for office. In-house Technology Counsel to the Commonwealth of Massachusetts (which by the way he notes would be a Fortune 50 company if it were private). Standards architect. Stanford researcher. Platform builder. He went to law school, he tells me, because he kept getting different answers from different lawyers to the same question and found it unacceptable. After years of practice, he still does not have a fully satisfactory answer to “what is the law?” but he at least knows how to find the relevant law himself, which was enough to let him return to technology without feeling incomplete. He was doing legal tech, he says, before anyone called it that. Writing scripts to automate his work. Treating legal documents as data - to the extreme displeasure of colleagues who just wanted Microsoft Word (why is it always Word?) Why didn’t the standard stick? In the early 2000s, Dazza was one of the architects of LegalXML, an effort to create an international standard for marking up legal documents so they could be treated as structured data. He ran the e-contracts group. It took seven years to reach the status of a recognised international standard. It attracted a small community of who he describes as lawyer-geeks who marked up their contracts in XML, built clause libraries, and imagined a future of genuinely interoperable electronic contracting. It did not really arrive, at least not in the way the group expected. A handful of vendors adopted the standard, mostly for workflows they were already running. The broader transformation never came. The lesson Dazza draws from it is that the obstacle was never the standard. It was the culture. The love affair with Microsoft Word was not something a well-designed schema could fix. Standards, he says, have to arrive at the right moment. Too early and the industry is not ready. Too late and people have already locked into whatever is there. The good news is, the moment has now arrived, and it came from a different direction entirely. Large language models can peer into the meaning of a legal instrument and address it as data without anyone having to tag a single element. Lawyers, it turns out, are naturally good at the lingua franca of LLMs: precise language, conditional logic, sub-clauses, if-but-not-that constructions. The standard that nobody could agree on turned out to be language itself. Is your agent loyal to you? Dazza has just finished a research sprint at Stanford’s Digital Economy Lab on a project called Loyal Agents, run jointly with the Consumer Reports Innovation Lab. The question it asks is simple and the implications are not: when an AI agent conducts transactions on your behalf, what legal framework governs whether it is actually acting in your interest? His answer keeps returning to fiduciary duty, and specifically to a US federal case called Kovel that almost nobody in legal AI is talking about. Kovel itself is sixty years old. It involved an accountant working with a tax law firm whose communications with a client became the subject of a grand jury subpoena. The Second Circuit held that the privilege extended to the accountant because he was acting as the lawyer’s agent in providing legal advice. The principle that emerges, Dazza argues, applies directly to modern AI vendors. To protect attorney-client privilege when a SaaS provider is handling client communications on behalf of a lawyer, that provider needs to be the lawyer’s agent in the legal sense. Most enterprise AI contracts disclaim exactly that. Dazza has read them all. He has built a public breakdown of what each of the major frontier model providers actually says in their enterprise terms about confidentiality and agency. See Links below. The most upvoted question at Anthropic’s recent legal webinar, which drew over 20,000 registrations, was about how to handle privilege when using general purpose AI tools. Dazza thinks the profession is looking for the answer in the wrong places. The technical controls matter. Zero data retention matters. But the legal layer, the contract clause that says we are your agent, is what Kovel’s logic requires, and most providers do not offer it. His pitch to the frontier model providers is direct: stop disclaiming agency and start defining it. Write a narrow, limited agency clause. Be your client’s agent for three specific things and disclaim it for everything else. It contains risk rather than expanding it, it supports privilege, and for providers building agentic products, it simply reflects reality. If a human did what these systems do, it would be an agent. In Dazza’s opinion, the contract should say so. The platform nobody has built yet Six months ago, Dazza started building something called Interlateral. The felt need behind it is something he has observed sitting in meetings in San Francisco with startup and innovation teams where everyone has agents running quietly in the background, and then communicating with each other through the narrow human-to-human channels of Slack and email, as if the agents are not there. Interlateral is a shared collaboration space where humans and their agents can work together in the same room. You bring your agents. He brings his. There is a third space where they can interact, collaborate, and co-work, with a shared markdown surface that both humans and agents can read and write into. The design principle is human-centred: a person is always at the wheel. The agents are extended cognition, not autonomous actors. The first event ran at Stanford last week with 60 lawyers and eight teams, and the next is at MIT. Eventually, Dazza wants tens of thousands of participants. He thinks the combination of people and agents in a shared space is a genuinely new source of collective intelligence, and that we have barely started to understand what it can produce. There is a harder problem underneath it. In Google Docs or Slack, identity is straightforward. You can see who wrote what. In a space where agents are acting on behalf of humans, you now have two levels of separation from the person you think you are dealing with. Agent identity and attribution, knowing whose agent it is and holding humans accountable for what their agents do, is a bleeding-edge question the industry has not yet solved. How he builds with agents At the end of our conversation, Dazza pulls up his GitHub repo and walks me through how he actually works. The interlateral_agents repo is open source, the product of years of slow tuning, and it is more architecturally interesting than most people’s agent setups. He runs three models in parallel: Claude Code, Codex, and Gemini, with Grok CLI from xAI expected to join shortly. What makes it unusual is the communications layer. The agents share what he calls a multiplexing comms hub, a setup that lets them read each other’s outputs and write into each other’s terminals directly. He describes it as a Vulcan mind meld. One agent can see that another tried something, that it failed, and suggest an alternative approach. The collaboration is explicit and adaptive rather than just running tasks in sequence. On top of that he uses skills: lightweight prompt-level definitions that tell the agents how to organise their work on a given task. He can arrange them hierarchically, with one agent orchestrating the others, or as peers collaborating on the same problem. The skills determine the shape of the collaboration without requiring complex infrastructure to enforce it. It is, he says, a surprisingly low-key way to get a lot out of very capable but very different models working together. What is the AI-native organisation? I ask Dazza what he thinks people are underestimating. The current pattern, he says, is that AI creates extraordinary efficiency at discrete points in a workflow and then causes congestion at the parts downstream that have not changed. Contract review is faster. The humans waiting for the output of contract review are not. The clog forms between the transformed part and the untransformed part, and it is going to get worse before anyone fixes it. The AI-native organisation is one that has redesigned itself around AI, touching pricing models, staffing, role definitions, and quality control, which starts to look less like a periodic event and more like continuous monitoring. That redesign, he says, is not premature. It is coming whether organisations are ready for it or not. The ones doing the mapping exercise now, looking holistically at the full lifecycle of a matter rather than optimising individual tasks, are the ones who will navigate it gracefully. The ones waiting are storing up a serious problem. Final note Dazza Greenwood is genuinely hard to keep track of. In preparing for this conversation I found Stanford research, open source repositories, consulting work, a platform under active development. Dazza literally switched hats midway through our discussion. What surprises me most though is that none of it feels scattered. The dots (and hats) are actually very connected. It all connects back to the same conviction he has held since the start: that law and technology are not separate domains, that legal instruments are data, that agents will conduct transactions on behalf of humans and the frameworks governing that need to be built carefully. He has been pretty patient about this for forty years. The future, he says, has finally arrived. And he seems genuinely delighted about it. Links * dazzagreenwood.com [https://dazzagreenwood.com], Dazza’s blog including his upcoming open source model comparison table * computationallaw.org [https://computationallaw.org], Dazza’s write-up of the Stanford Interlateral event * interlateral.com [https://interlateral.com], sign up to attend a future event * civics.com [https://www.civics.com/]to understand more about Dazza’s consulting work * loyalagents.org [https://loyalagents.org], the Stanford and Consumer Reports research and vendor contract analysis, including the Kovel breakdown * analysis on when agency is a feature not a bug [https://loyalagents.github.io/loyal-agent-evals/report/#2-3-2-when-agency-is-a-feature-not-a-bug-the-kovel-pattern-in-legal-tech-saas] * github.com/dazzaji/interlateral_agents [https://github.com/dazzaji/interlateral_agents], the open source agent library Dazza uses to build with Claude Code, Codex, and Gemini The Technical Stuff Here’s a quick primer on some concepts that may be unfamiliar to some: Evals Short for evaluations. A structured way of testing whether an AI system is doing what you want it to do, consistently and measurably. Dazza uses an open source platform he built to run evals on agent behaviour, putting numbers on whether an agent is acting in a user’s interest or getting tripped up by a conflict it has not recognised. Think of it as quality control, but for AI decision-making. Fiduciary duty A legal obligation to act in someone else’s best interest rather than your own. It applies to lawyers, financial advisors, and other professionals. Dazza’s Loyal Agents research asks whether AI agents should be held to something similar, and whether the contracts governing AI services currently reflect that expectation. Most do not. Kovel A 1961 Second Circuit case that Dazza thinks is the missing piece of most privilege discussions in legal AI. United States v. Kovel involved an accountant employed by a tax law firm whose communications with a client became the subject of a grand jury subpoena. The court held that the attorney-client privilege extended to the accountant because he was acting as the lawyer’s agent in providing legal advice. The principle Dazza extracts: for privilege to hold when a SaaS provider handles client communications on a lawyer’s behalf, that provider needs to be the lawyer’s agent in the same legal sense. Most AI vendor contracts disclaim agency explicitly. Dazza argues this is both a legal risk and a fixable problem, and that providers who address it will have a commercial advantage in the legal market. Multiplexing A communications technique that allows multiple signals to share the same channel. In the context of Dazza’s agent setup, it means his agents can read each other’s outputs and write into each other’s terminals in real time, rather than operating in isolation. The result is agents that can observe what the others are doing, flag when something is not working, and adapt accordingly. Skills In the context of agent configuration, skills are lightweight prompt-level instructions that define how an agent should approach a task or how a group of agents should organise their work together. They are not code in the traditional sense. They sit closer to a well-written practice note. Anthropic has its own Skills framework, which we have covered separately on agents.law. Loyal Agents Both the name of the joint Stanford Digital Economy Lab and Consumer Reports Innovation Lab research project Dazza works on, and a broader concept: the idea that AI agents conducting transactions on behalf of users should be demonstrably aligned with those users’ interests, in a way that is measurable, contractually grounded, and legally enforceable. The research has produced evals for testing agent loyalty and a public analysis of how major AI providers currently handle the relevant contract terms. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.agents.law [https://www.agents.law?utm_medium=podcast&utm_campaign=CTA_1]

Should Law Firms Buy or Build?

Michael Kennedy, Addleshaw Goddard Michael Kennedy left university swearing off being a lawyer. He went to work in restaurants and retail for a while, and finally came back to it as a paralegal at Addleshaw Goddard. When he started his training contract, it was very much focused on innovation and technology - a rarity at the time, and Mike was one of the first in the UK to follow this path. Since then, the innovation team at AG has grown from a handful of people to around 80 today. Fast forward to 2026, and Mike now runs the firm’s R&D function, a broad role that encompasses horizon scanning, startup engagement, partnering with clients, internal education, leading a development team, and increasingly a fair amount of building things himself. How do you project into the future? I asked Mike how anyone keeps up when the world is spinning this fast. His answer is that he doesn’t really switch off. He reads constantly, runs research agents through Claude Code, and writes a fortnightly internal newsletter for his team: three things to know, three deep dives, then a long reading list. He says it’s really important for him to know what he’s talking about. He’s not afraid to say he doesn’t know, but he doesn’t like saying it, so he’d rather know and do it. AGPT and the case for building The most visible output of Mike’s team is AGPT, the firm’s in-house AI tool. It’s one of the rare examples of a law firm having built its own in-house solution at a time when most of the industry is focused on buying legal AI. Mike describes AGPT, with characteristic understatement, as “what most people just call a wrapper”. It lives in the firm’s Microsoft Azure environment and does the usual things: chat, document review, translation, prompt libraries, citation tracking. In early 2023, Mike’s team wanted a sandbox to test whether GPT-3.5 was good enough for legal documents. They couldn’t throw client matters into ChatGPT, so they asked the developers to stand something up inside the firm. Other lawyers started asking for access. A pilot followed, then a firm-wide rollout by autumn 2023. The sandbox became the product. Today AGPT runs around 6,000 prompts a day across roughly a thousand users, and the dev team hasn’t had a quiet week since. Mike’s buy-versus-build framework is worth listening to because it comes from someone who has actually done both. Cost matters, but he frames it as a return-on-investment question rather than a sticker-price one. The real factors, he says, are: * Institutional knowledge: you can build for your specific audience in a way you can’t buy for one. A product on the market might have 70% irrelevant features and lose people before they engage, whereas a 30% solution built for your lawyers can land better. * Client consent and data: self-hosted removes a lot of friction. * Portability: which he thinks is underrated. The value is in the solution to the problem, not the tech it happens to run on. In Mike’s view, law firms are going to want to move their prompt libraries, workflows and accumulated know-how between models, and the firms that treat their intelligence as tech-agnostic will have an easier time than those locked into a single vendor. He uses a nice phrase for this: portable intelligence. Claude, vibecoding, and the artifact economy Mike and I are both heavy Claude users. He uses Claude Code primarily to build prototypes - someone in the firm has an idea, usually hard to execute, and instead of taking notes and going away for six months, Mike builds a rough version and shows it to them. One recent example is a regulatory horizon scanning tool for banks, the kind of thing his financial regs team has wanted for a long time. Not a finished product but enough to say “is this what you mean?” and have a real conversation with the Partner. On Claude for legal work itself, Mike is bullish in a way that he thinks should worry the established legal AI vendors. A lot of work inside law firms isn’t legal research. It’s factual research, web searching, document comparison, content creation, the work that fills a competition team’s afternoon. That work is dramatically easier in Claude than in Google, and Mike says Addleshaw is, for the first time, seriously considering enterprise licenses for the lawyers and teams who would benefit. Thanks for reading Agents.law! This post is public so feel free to share it. Training junior lawyers in a world with fewer trainee tasks The question that won’t go away is what happens to junior lawyer training when the grunt work disappears. He thinks training in law firms has always sort of worked by accident. Trainees are bright, engaged, hard-working people who pick things up by osmosis, sat next to a supervisor with a red pen. It’s slow, it’s inconsistent, and what you actually learn is often one individual’s approach rather than a structured body of knowledge. His proposed solution is a good one: use the firm’s data and know-how to build anonymised simulations based on real client matters. Give trainees scoring, measurement, and a structured way to develop across different areas. If AI reduces the billable work trainees do by 20%, use that time for simulated exercises rather than cutting trainee numbers by 20%. It’s an optimistic framing, and Mike knows it. The realist’s version is that firms will just fill the hours with more work and make more money, because that’s what the economic incentives reward. Side note: Mike has built a prototype for this - it’s live on my Vibecode.law platform [https://www.vibecode.law] - check it out! Final note What I take from my conversation with Mike is that there’s a particular kind of legal innovator becoming more common in the industry. They’re not pure technologists and they’re not pure lawyers. They’ve got enough technical ability to build, enough legal experience to know what matters, and enough organisational patience to sit inside a big firm and make things happen. In my view, a lot of the interesting change in BigLaw over the next few years is going to come from “intrapreneurs” like Mike, inside firms, building things and encouraging others to do so. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.agents.law [https://www.agents.law?utm_medium=podcast&utm_campaign=CTA_1]