The AI Samurai: building the ‘AAA’ formula with Sara Maldon, Make

John Emmerson: Sara, thank you for joining me today. You are Head of Business Automation and AI at Make, and you’ve been leading something called Miriwa, which is your internal AI transformation programme. People were probably already using AI tools informally before Miriwa launched. Why did you feel it needed a formal programme on top of that?

Sara Maldon: I was actually hired to lead this formal programme. There was already a recognition of a need before I joined — I joined specifically for this purpose. At the time, two years ago, there was no official AI tool. Everything was prohibited. There was no strong sense of vision around where we were going with AI. But there was a strong understanding that this was extremely important for an organisation like Make. Not just because of efficiency and how we reinvent the way we work internally, but more so because we do offer an automation platform, and AI fits naturally with that. So it was about how do we do this right, how do we experience that internally ourselves first, before anyone else figures it out?

John Emmerson: Do you know what leadership needed to see before they greenlit the programme?

Sara Maldon: We started the investment a little differently than most companies. The conversation was: we don’t know what we can achieve, we don’t know what we don’t know in the area of AI, but we know this is going to be very important. So we need someone to help us figure it out. I was the person hired to do that — which meant there wasn’t necessarily a defined problem to solve yet. There was an existing automation culture at Make: do it once, do it twice, automate it, never do it again. That culture was already there. What they were really looking for was someone to put a vision in place and execute on it, so that people didn’t just feel supported through the change but felt empowered by it, in a way that helped their careers.

John Emmerson: When you started two years ago, AI adoption within the business was at zero per cent. And now it sits at 96 per cent. That’s not just a tool rollout — that’s a whole cultural shift. What was your biggest obstacle?

Sara Maldon: The difficult thing about an AI transformation is that you always have a moving target. You start with an AI tool — ChatGPT, Gemini, Copilot, whatever your organisation uses. You introduce it, start with adoption, then start figuring out use cases, get people excited. The moment they engage, you start introducing AI automation. The moment you understand AI automation, you start introducing AI agents. And the field is moving very quickly as well. Whatever you do in AI transformation, the challenge is that you need to embrace that it’s a moving target. We have 96 per cent of people building AI automations — roughly half are agents, half are AI workflows. But now we want to see people sharing their most impactful AI agents. So we’re moving that bar to the next step.

Sara Maldon: You can’t do AI without an AI tool. That’s where it starts. Once you get people excited about adopting AI, the biggest challenge is making sure they have time for it, making sure they’re excited, making sure the right resources and enablement are in place, making sure managers are enabled and lead by example. You need top-down buy-in to get the bottom-up buy-in. And then when people start building, you need to make sure they know what to build and that it’s meaningful. There’s always a challenge at every step. But it’s also probably the most meaningful thing to be working on in an organisation right now, because you can see how it changes the way people think and operate.

John Emmerson: What’s the silver bullet? Was there one?

Sara Maldon: I’m always very cautious about silver bullets. There is a lot of hard, hard work and a lot of investment from a lot of people behind it. When I was hired two years ago, I didn’t have a team. It was about figuring out a vision. I got a part-time resource to help me — a really excellent person to help test and implement some projects. We went into the most strategic but also the lowest-hanging fruit. Small effort, big win types of situations. We implemented those first and that helped us prove the business case for AI implementation. We did a few trainings at the beginning. In the first week of my role, I organised a four-week bootcamp and a third of the company signed up. They were really excited to understand what was happening in the market.

Sara Maldon: Then it changed. Once we’d validated the business case, it shifted from a curiosity effort to an intentional effort to transform the organisation. My reporting lines changed, I got a dedicated team for every single department. We call them the Samurai — the people in my team who helped make this a reality. They’re full-time people dedicated to this, one in HR, one in engineering, one in product, one in marketing. They also report to the VPs of those departments, so there’s real accountability. The shift was: start somewhere small where you can prove the effort and build confidence in both leadership and employees. Then go from there with intention. Make sure it’s funded, make sure it’s on top of the agenda, and then execute.

John Emmerson: You’ve got 96 per cent adoption. What about the four per cent who haven’t engaged?

Sara Maldon: We’re investigating whether that’s people who just didn’t want to do it, or people who missed their deadline, or people who didn’t find the right use case, or who were simply new joiners. But I’d highlight: it’s never been about the four per cent. It has never been about that. We have a saying: focus on the one who wants, and invest the energy there, because that’s where the success will be. Our focus now is on the impact of agents and on vibe coding — making sure we’re not just focusing on AI automation, but really on the agents and on building structure around that.

John Emmerson: Miriwa is built on what you call the AAA formula — Automation, AI and AI Agents. How did you land on that structure?

Sara Maldon: You’re going to laugh, but it used to be a double A and then people were saying it sounded like Alcoholics Anonymous. So that was a branding issue. We started from our DNA — we started with a single A, automation. We had that mentality already: do it once, do it twice, automate it, never do it again. We added AI when I joined, so we had AI and automation culture. And then in April last year, that’s when we launched Agents and we rebranded to Automation, AI and Agents. Right now, agents are particularly important for client-facing roles, because they’re new and these teams haven’t engaged with them yet. For technical implementation functions, we ask them to think about what’s the right solution and build for impact, because very often people start with agents and then realise that a standard automation is actually much more meaningful for that particular problem.

John Emmerson: What do the best Samurai do that the others don’t?

Sara Maldon: There are two very strong skill sets that make a really good Samurai. The first is curiosity — and curiosity means you will keep moving your benchmark. When the latest agents were launched, the best Samurai would have gone and tested them out immediately, even built a first application, gone and figured out the edges, and then shown the rest of the organisation what they learned. They wouldn’t wait for someone to tell them this is good or to build it out and show them how to do it. The other skill is this people-centric ability to understand what a problem is, even when it’s hidden between the lines, and then obsess about it and build the right solution — not from a technical standpoint, but the right solution for the person in front of you. That’s empathy. It’s almost a product manager role. You’re obsessing about a problem and figuring out with empathy what will work for that particular user and how you can bring them through the change.

John Emmerson: Early on, quite a few of your AI projects failed. But more recently you shipped 23 successful projects out of 32. What changed?

Sara Maldon: It’s very normal when you start. We track the success rate of projects, which for us means: did this project achieve the impact it set out to achieve, and is that impact on the right metrics? I don’t recognise time saved as the right metric. If a project achieved some time saved, I don’t recognise it as a successful project. When we talk about real impact, we talk about core business metrics. It’s also very natural at the beginning that most projects have a low success rate because the team is new and they’re new in their role — a role that hadn’t existed before. And the quickest way you learn is through failing. Always.

Sara Maldon: And you have to do that early, as early as possible, because then the anticipation of perfection keeps building and you have more and more to lose. There was a fun conversation I had with someone after they’d been onboarded — it had been seven weeks and they hadn’t failed yet. I went to them and said: I want to see a failure next week. Otherwise I will throw you into one.

John Emmerson: On infrastructure: you’ve written that agents can amplify what’s already there — if the foundation is solid they become a multiplier, but if it’s messy they amplify the mess. What does getting the foundation right actually look like?

Sara Maldon: I also meet a lot of businesses that overthink this. They always say: we don’t have the right foundation, so we can’t build this or that. There’s always a small win you can have that can be quite impactful. But if I tackle the foundation question directly: sometimes you start with an agent too early, because your executives come to you and say here’s an AI budget, agents are a big thing, go figure things out. And you need to show them you have an agent. But actually what you realise is that you first need the data to flow from A to B to C in an automated way — you need a process. Then you can build intelligence on top of that process, which is what an agent does. There’s always a way to look at this as an opportunity. You can say: this is not for us yet. Or you can say: I need an agent, which means I need to be advanced in something — so let’s get cracking and figure out the foundations within the scope you need, and still keep the agent as the ultimate goal.

John Emmerson: How do you spot curiosity and courage in an interview?

Sara Maldon: If you ask one of the Samurai, the first thing they’ll tell you is that I do ask very particular questions. It’s actually a running joke — they all remember the questions I asked because they always had to pause and think. For me, it has always been about how someone approaches problems and what they focus on. It’s almost like a product management role. You need to understand the people, know the people, have their trust, be in love with the problem, be obsessed about the problem. Once you do that, then you need a technical skill. The technical skill is more of a tick-box. I can spot immediately whether someone has the technical skill. But how did you think about these choices? Why did you choose this solution? What do you think would be the first reaction of a person using this? I’m seeing whether they’re considering not just the people they’re building for, but also how it impacts others.

John Emmerson: Back in July 2025, when you announced Miriwa, you were targeting 90 per cent AI literacy and 75 per cent of departments executing real projects. How did you perform against those goals?

Sara Maldon: I’m very bullish on setting ambitious metrics — ambitious enough that they feel within reach, but where it’s unlikely you’ll fully reach them. That’s how you set a metric. In a high-performance culture, you need what we call big, fat, hairy goals. For AI literacy we almost reached it — four per cent short. For the 75 per cent implementation target, we ended at 73 per cent, so within two per cent. Which means we have to raise the goals for this year.

John Emmerson: Has Miriwa made Make measurably more efficient, faster at shipping, better at competing?

Sara Maldon: The biggest impact we’ve had: revenue per FTE went up. And general revenue and sales metrics, particularly go-to-market metrics, were also particularly successful. AI and automation is especially impactful in go-to-market functions — you would expect that if you invest in implementing AI and automation in marketing and sales, it shows in revenue. One of the metrics we were also focused on was time-to-fill in hiring, because if you hire the right people in the right place at the right time and do it fast, your organisation will also massively speed up.

John Emmerson: What’s the thing that doesn’t get talked about enough that determines whether all of this works?

Sara Maldon: The level of belief you have in it. If you’re hiring for people who are courageous and curious, you need to be that yourself. You need to be ready. Once you say you want to do this, you need to mean it. If you don’t mean it, you’re not going to be successful. AI transformation is difficult. Your leadership needs to provide the resources, your IT needs to be ready because it’s a tooling question at the start, your HR needs to be on board because it is a human question at its core. If you want to do a cultural transformation, you need your Chief People Officer on board, excited, and potentially the first person to build something. The leadership needs to be there. You need to mean it. That’s the last bit.

What is Miriwa and what was it designed to achieve?

Miriwa is Make’s internal AI transformation programme, designed to move the entire global workforce from zero AI tool usage to active AI implementation. Sara Maldon was hired specifically to design and lead the programme, starting from no formal AI tool access and building towards measurable business outcomes. The programme is structured around Make’s AAA formula: Automation, AI, and AI Agents.

How did Make achieve 96 per cent AI adoption across its workforce?

Make’s adoption rate was built through a combination of early curiosity programmes — including a voluntary four-week bootcamp where a third of the company signed up — followed by an intentional, resourced transformation effort with dedicated AI Samurai embedded in every department. The Samurai are full-time specialists who serve as coaching partners, thought partners and implementation guides for each department. Leadership buy-in at VP level ensured genuine accountability. Sara Maldon is clear that there is no silver bullet: the results came from hard work and sustained investment over two years.

What is the AAA formula for AI adoption?

The AAA formula stands for Automation, AI, and AI Agents. Make began with an automation culture, added AI (intelligence and reasoning layered on top of processes), and then added AI Agents (autonomous AI systems that can take actions and make decisions within defined parameters). The framework helps organisations understand where they are in their AI maturity and what the right next step is, rather than jumping straight to agents before the foundations are in place.

What makes a good AI Samurai?

Sara Maldon identifies two core traits: curiosity and empathy. Curiosity means continuously testing new capabilities, pushing at the edges of what’s possible, and not waiting to be told what to explore. Empathy means understanding the human problem behind a request, not just the technical specification. The Samurai role is essentially a product management role applied to AI implementation — obsessing about the problem from the perspective of the person experiencing it, and then finding the right solution for them specifically.

How should organisations think about failure in AI projects?

Sara Maldon’s view is that early failure is essential, not a problem. Teams that are new to a role will naturally have a low success rate at first, and the quickest way to learn is by failing early, when the stakes are low. She explicitly told a new team member who had not failed within their first seven weeks that she expected to see a failure the following week. The anticipation of perfection, she argues, is the silent killer of AI adoption.

What does getting AI foundations right actually look like before deploying agents?

Before deploying AI agents, an organisation needs data to flow reliably from one system to another through automated processes. Agents work by building intelligence on top of existing processes — if the processes are not in place, agents amplify the mess rather than the value. Sara Maldon’s practical advice is not to treat missing foundations as a blocker, but to use the desire for an agent as the motivation to build those foundations.

How should organisations measure AI success?

Sara Maldon rejects time saved as a primary metric. In her view, time saved is a phantom KPI if it does not translate to movement on core business metrics — revenue per FTE, time-to-fill, sales and go-to-market performance. If a project achieved some time savings but had no impact on the metrics that matter to the board, she does not classify it as a successful project. Organisations should define what success looks like in terms of business outcomes before they start, and then measure against those outcomes, not against usage statistics.