AI and the Innovation Economy with Ashley Williams

Jack: Thanks for joining me today Ashley! We’re here at your headquarters in London, England. Why don’t we start out with you telling me a little bit about yourself and Mishcon?

Ashley: I’m the Head of Technology Group here at Mishcon, and I also lead our AI sector group. For those that don’t know Mishcon, we are a law firm that focuses on supporting innovative clients within the innovation economy — all the way from really early-stage ideation companies through to international conglomerates and everything in between along their journey.

Jack: So you probably get to see which industries are being most impacted by AI today. Can you walk through what you’re seeing at the frontier — which industries are being most profoundly impacted by AI?

Ashley: We’re very fortunate to have that helicopter view across a lot of different sectors. I would start with any that are heavily content-driven. From the launch of generative AI and the mainstream explosion we saw in March 2023, those have seen the most change, and the quickest change. We spent a lot of time working with marketing companies figuring out how to change their business models and get ahead of generative AI.

But maybe the more exciting change — and I would say this partly because we act for one of the biggest players in this market — is around autonomous vehicles. Seeing the pace at which we have moved from traditional vehicles to cars on the road that actually drive themselves has been incredible.

Jack: This is the most physical manifestation of where AI is taking us, and it’s a bit unnerving. I was in San Francisco recently, and there are oceans of autonomous vehicles driving people around. It’s just a little jarring the first time you see it. So that’s on the cusp of arriving in the UK?

Ashley: We’ve been running pilots for a while, but we’re quite excited by the next phase, which is autonomous passenger services. What some people call “robotaxis” is the next big thing in the UK. And I’ve had the joy of being in the back of one of those vehicles — even seeing someone in the driver’s seat but not touching the wheel takes your brain a little while to catch up.

Jack: It takes a beat for you to process what’s happening.

Ashley: You almost want to lean forward and make sure someone is there. But it’s crazy how quickly your mind adjusts — after just ten minutes in the seat, it starts to feel normal.

Jack: That was my experience in a Waymo in the Bay Area. The first ten minutes I was videotaping, sending clips to family, raving about it. But 20 minutes in, I’d essentially forgotten I was in a driverless car. It’s amazing how quickly you go from something that feels radical and uncomfortable to just background noise.

Ashley: Trust is a huge part of this. What excites me about passenger services is that it becomes mainstream — lots of people will have that experience in the UK, and you start to build public trust in autonomous vehicles and AI in travel. That’s one of the hardest things in this sector: making sure we bring people along on that journey so they start to trust AI in everyday activities.

Industries & the Data Challenge

Jack: Beyond autonomous vehicles, what’s following that wave in terms of AI adoption and transformation?

Ashley: There’s a headache for a lot of companies around benefiting from AI, and it often comes down to data. Many companies are data-rich, but the ability to extract that data into a format AI solutions can actually consume isn’t always there. A lot of the work we’re seeing right now is getting data into a consumable format — and once you do that, you unlock enormous potential advantages.

A really obvious example is libraries. They are incredibly data-rich, but a lot of it is in physical or book form. Imagine a world where all of that was OCR’d overnight and you suddenly have millions or billions of data points accessible to AI. That’s exciting.

Jack: So one challenge is wrestling with this huge quantity of data and structuring it in a way that’s consumable by AI?

Ashley: Yes. And there’s another wrinkle: by using AI, we’re actually driving more content into the market as well. So the question becomes, what is good data and what is bad data? When we talk about AI companies, it really is all about data — how we manage it, how we get the best use of it, and how we make sure we don’t ingest data that will compromise the models.

Challenges for AI Companies

Jack: What are the primary challenges you see AI companies trying to navigate — and maybe some challenges they don’t anticipate?

Ashley: There are some obvious commercial issues. Compute is getting harder and harder to come by. The deals we’re doing in that space aren’t million-pound deals, they’re billion-pound deals of committed compute.

The other piece is making sure you’re solving a real problem. A lot of the early-stage startups we act for end up pivoting along the way, because what they thought they were solving for isn’t actually the problem people have.

And in terms of what’s coming down the line — more regulation. We’re seeing more in the US, and the EU has been on top of this for a long time. Navigating a whole new regulatory layer is something many AI companies have not been used to dealing with yet.

Jack: How do AI companies in the UK think about the competitive landscape?

Ashley: Everywhere. Competition is fierce, and the speed at which people are iterating and bringing things to market is unprecedented. None of our clients are domestic in mindset — everyone is looking further afield. The speed at which competition is coming out of the US, China, the Middle East — it’s just everywhere.

I think there’s an understanding that you operate in one of two ways: you’re the Swiss Army knife — everything to everyone, like the Microsofts of the world — or you go narrow and solve for a specific use case. Where we’re supporting most of our clients is in that focused, “let’s solve a real, solvable problem” space, and that’s where they’re having their most immediate wins.

Jack: How are your clients navigating the compute challenge?

Ashley: Some are turning to hyperscalers who offer on-demand, scalable resources rather than just committed capacity. We’re also seeing groups form together — creating shared stacks they can build and train on, and benefiting from economies of scale when negotiating collectively. A lot of partnerships are emerging, sometimes looking further afield globally for better compute deals.

What Makes AI Truly Transformative

Jack: What differentiates companies that are really embracing AI in a transformative way?

Ashley: I think it’s a goal beyond immediate profit. The most compelling companies have a goal in mind that is about leaving a legacy and moving the dial for the world. A great example is companies using AI for drug discovery — the power and speed at which we can now bring new drugs to market is unprecedented. Those companies that think beyond their own bottom line are where we’re seeing the most monumental impact.

Jack: DeepMind here in the UK is a great example — the breakthroughs in protein folding coming from an AI group.

Ashley: A great example. And what they’re doing is getting out of their own way — being willing to partner with even competitors, to help and deliver something in a different way than we’ve seen before.

The Hype Cycle

Jack: Where do you think we are in the AI hype cycle right now — heading into a trough of disillusionment, or already seeing real-world impact?

Ashley: I’ll disclose upfront that I’m a big believer in AI. But I think we’re not in an NFT-style bubble — that’s for sure. There are certainly some AI companies I’d call “thin skins” — companies just wrapping themselves around LLMs — and that will create some disillusionment. But we’ve just given great examples of where AI is having a real impact on world-changing problems.

Expectations and reality will always be slightly out of kilter, but the gap between the two now is much closer than it was five or ten years ago. I don’t think we’re about to hit a massive trough. There will be some recalibration, but by and large we’re heading in the right direction.

Staying Ahead of Frontier Models

Jack: Are the companies you’re advising worried about staying ahead of the frontier models? How do you build long-term differentiated value without getting steamrolled?

Ashley: People have learned from history. One of the things companies are building in by design is the ability to switch — not becoming completely dependent on a single frontier model, so they don’t have a single point of failure if one provider completely changes its business model.

That helps on two fronts: you don’t become dependent, and you can move quickly to benefit from advancements in new iterations. It also helps on cost — not all frontier models are equal on pricing, and you may not need the full capability of the latest model for every activity. We’re seeing companies say, “this activity requires less accuracy, we can use a lighter model; this one needs the latest frontier model.” Startups and high-growth companies are very acutely aware of the single-provider risk.

AI Regulation

Jack: How do you think about striking the right balance in AI regulation — between a dangerous free-for-all on one end and over-regulation on the other?

Ashley: I’m very pro-AI and pro-innovation, but I’m also pro-innovative regulation. What I mean is regulation that gives clear boundaries and clarity about where you can play comfortably, without preventing people from doing things safely.

Every UK regulator will immediately use the phrase “pro-innovation regulation.” But we also have an example of where it went wrong, in the EU AI Act. They were ambitious about wrapping real rules around AI usage, but perhaps went too far, too quickly before fully understanding the technology. A perfect illustration: just as the Act was about to be enacted, the generative AI explosion of March 2023 hit, and they had to go back and readjust the legislation in real time. That shows it wasn’t flexible enough to allow for advancements.

There’s a middle ground — less rule-based and more principles-focused, to allow for flexibility. But we absolutely need guardrails, especially when AI has an impact on the physical world and human safety.

Jack: What did the EU AI Act get right, and what did it get wrong?

Ashley: The principles are correct — transparency, explainability, accountability, governance. Those are the right topics, and we see them across many jurisdictions worldwide.

Where it’s fallen down is in the high-risk category. If you’re using AI even in recruitment — which is classified as high-risk — your obligations become very intensive and documentation-heavy. Sometimes it feels like documentation for documentation’s sake. And in recent months, we are seeing some of that regulation being dialled down.

Jack: What would pro-innovation AI regulation actually look like?

Ashley: Accountability is a big one — having governance in your organisation to understand who is responsible for AI safety. Explainability and transparency too: being clear and open with people that you’re using AI tools, especially where those tools may impact people’s employment prospects.

One area where I do support the EU Act is around data provenance — particularly for training models. A lot of companies have historically just scraped broadly to power their models without really understanding the provenance — and that has a significant impact on things like bias and discrimination in AI outputs.

Data Sovereignty

Jack: What should companies be thinking about when it comes to data sovereignty?

Ashley: The main thing to note is that some of the legislation around data localisation isn’t as difficult as it’s sometimes interpreted. Even where you have strict data localisation laws, it doesn’t mean you can’t still have global infrastructure supporting the delivery layer.

There’s definitely more partnership activity, more JVs, more combining of functionality with local expertise. That’s actually a positive outcome — it brings investment to local jurisdictions and helps them develop compute capacity domestically.

Jack: For a startup operating across multiple countries, how should they think about navigating different regulatory regimes?

Ashley: Our approach is to identify the high watermark. When you know the most demanding regime, you can apply those rules and get a passport effect into other countries. The EU GDPR on data is the high watermark — if you understand compliance around that, I typically tell clients you get an 80/20 passport into other countries.

But it’s not just the legal framework — it’s also about customer demand. You may comply perfectly with local regulations and still miss a major issue around how customers in that jurisdiction expect data to be managed. You cannot underestimate the value of local knowledge.

AI at Mishcon

Jack: How has AI transformed how Mishcon delivers services to clients?

Ashley: Mishcon has never been shy about innovating in service delivery — it’s always been in our fabric. Every client is on their own AI journey, with their own tools and way of thinking about this. So when we develop internal tools, we have to be tech-agnostic on the client side — we want to meet clients wherever they are.

What we’re increasingly talking about is managed services. We see ourselves as having an important role in providing this next wave of services in a way that brings efficiency and speed — but more importantly, frees up our time to get to know clients better and do real horizon scanning. Not just dealing with today’s need, but predicting what their issues will be in five or ten years time.

And we’re doing things with these tools now that were simply impossible from a pure manpower perspective before. A great example: a client was walking into a board meeting and realised he needed a summary of a 300-page agreement no one had properly reviewed. He called in a panic. The response was, “we don’t need to send someone, we can turn this around in ten minutes.” The ability to summarise and extract — the tools are just exceptional.

Billing & Business Models

Jack: Have you evolved how you bill clients as a result of AI adoption?

Ashley: If I could get rid of the billable hour tomorrow, I would. It’s so archaic. The question when talking to clients about fees should be: what value are we actually going to add, and what certainty can we give around those fees? Fixed fees, capped fees, block hours — they’re already part of our fabric, but we’re still figuring out the best models for managed services going forward. What’s exciting is that our clients are very willing to go on that journey with us.

Jack: Clients have no affection for the billable hour.

Ashley: None. Their willingness to try something different is very high. Maybe we can push the envelope even further and talk about value-based pricing — what are we actually doing, where are we moving the dial, and what does that account for in our fees?

I’ve got another 30 years ahead of me. I don’t have a choice but to play around and figure this out. And our managed services rollout will involve no talk of hours or billable units. It’s going to be a completely different way of billing.

Legal’s Spreadsheet Moment

Jack: I think legal is going through its spreadsheet moment. When Visicalc arrived in the late 1970s, pundits predicted the accounting profession was finished. But accountants moved up the value chain — they became better partners to their clients. NDA analysis is the adding machine work that doesn’t need to be done by a human being anymore. The opportunity to become a true strategic partner to your client is greater than it’s ever been.

Ashley: You can see it in how junior lawyers respond. Ask a junior to manually review an NDA and the shoulders go up — “okay, another one.” But the conversation I’m having now is completely different: “Do you want to help devise an end-to-end solution for a client who processes a thousand NDAs a year from one department alone?” Hands go up pretty quickly.

Jack: On complex matters, AI lets you deliver a better work product more quickly. Arguably that should cost more, not less. What’s your take?

Ashley: I completely agree. What I can do now is take on massive disclosure projects that would previously have been a real stretch for my team, and deliver them at speed and at the quality standard clients come to us for. We could probably run three or four of those projects simultaneously now. Give us six to twelve months, and I expect our marketing team will be very excited to be talking about some of these wins.

Jack: The dynamic capacity AI gives you is really interesting — it’s not just one lawyer doing things faster, but potentially doing the work of four, five, ten lawyers.

Ashley: Absolutely. And for our international clients, we can now deliver training and education in 20 languages in real time — so a colleague in another country has the exact same experience as someone sitting in our London office. That is transformational in how we deliver knowledge to a global client base.

Career Journey

Jack: You’ve had an incredible journey to heading up Mishcon’s Global Technology Group. Tell me more about that story.

Ashley: I always knew I was going to get into technology — I was one of those nerdy kids who used to build computers with his uncle. And I always knew I wanted a vocational path, and law was an obvious pull. I trained at Addleshaw Goddard and loved it, but they weren’t doing the level of tech work I wanted.

I wanted every day to be almost entirely technology — serving tech clients. So I moved to an American firm where we served the likes of Google, Apple, and Facebook, which was brilliant. But I also felt like a small cog in a big wheel, not sure how much I was moving the dial.

From there, things became more purposeful. I wanted to get into the VC world, work with high-growth companies, and moved to a boutique tech law firm. That brought me to where we are now at Mishcon — same boutique tech mindset, but with the full capability of a large international firm to help companies go from ideation all the way to exit or IPO.

Jack: It sounds linear, but I’m sure it wasn’t. What were some of the real challenges along the way?

Ashley: The biggest moment was about four or five years ago when I genuinely thought I would leave law and flip to the business side. And then I found a boutique firm operating very differently — where some partners had developed into almost chief commercial officer roles and sat on boards. That changed my mindset back: I like being a lawyer, I like bringing that expertise, but I want this other dimension too. And maybe I could have both.

The greatest joy I get from my work is clients calling me with questions that are almost entirely non-legal — things like, “I’m about to do a channel partnership in America, ignoring the contract, what should I be thinking about?” Someone who’s seen that transaction three or four hundred times before can distil all of that knowledge into one conversation. That is a really good use of an hour.

Advice for Young Lawyers

Jack: If you were advising a young lawyer just entering the industry — wondering about the impact of AI, and how to see it as an opportunity rather than a threat — what would you tell them?

Ashley: Look at skill sets beyond the law. Go and understand how clients are actually building this technology. Do a coding course. Develop skills that will help you understand what the client’s real need is and what challenges they’re facing.

Lawyers who enjoy their subject matter and love what their clients are doing are the best lawyers, because they get deep into the detail of the client, the project, the product. If you’re opposite someone as a counterparty and you understand the technology, you understand the issues. We’re always operating in grey areas. You’ll feel much more comfortable making a decision in that grey area if you know how the product is built. You can give more value back to your client, and you can do it faster.

Jack: I think a lot about empathy. How do you put yourself in your client’s shoes and truly understand their problems? You could do a crash course with an AI tool over a weekend on virtually any discipline, and get that much closer to your clients and the real problems they’re facing.

Ashley: 100%. Become a prompt engineer — really understand that market and how it will develop, because you will absolutely need that skill set going forward.

We also spend a lot of time embedding our team with clients for six to twelve months, so they really understand what those clients are facing. And our hiring has changed significantly — we’re bringing in people who have been in-house at tech companies before joining us. If I have a team around me who have walked the walk as General Counsel, that conversation with clients is far more fruitful and authentic than someone who has never experienced an in-house environment giving advice.

Jack: This has been a fantastic conversation. Ashley, thanks so much for joining me today.

Ashley: My pleasure. Thank you for having me.