The Question Most Leaders Don't Ask: Why Liganova MaSH! Started With a Blank Sheet
By Polly Barnfield, OBE, CEO of Maybe*
Most AI deployments start with a tool. Liganova MaSH! started with a question. The difference is why they are in the 8% of businesses who have actually operationalised AI, while the rest are still wondering why adoption stalled.
Less than 12% of businesses are using AI in a meaningful way. Only around 8% have genuinely operationalised it. That is not a technology problem. It is a deployment problem. And the difference between the 8% who got there and the 92% who didn't almost always comes down to the question they asked at the start.
Most organisations ask: what can AI do for us? Someone in leadership sees a demo, buys licences, and asks the team to use the tool. Adoption stalls. The tool becomes another thing to log into. The C-suite reports saving 12 hours a week, while the rest of the workforce reports AI adding to their workload because they are spending time learning tools that do not fit how they actually work. That is not me speculating. That is what the Wall Street Journal found in its own research.
Liganova MaSH! asked a different question. Not what can AI do for us, but: what do we want our people to stop doing?
That is the question this edition of Scaling Up With Agents is about, because it is the question that determined everything that came after it.
Who Liganova MaSH! is, and why getting this wrong was not an option
Liganova MaSH! is the agency behind BMW's global social platforms and content operation. They produce over 2,500 pieces of content a year for BMW alone and reach nearly 100 million people across social. When you are running creative at that scale, under that level of brand scrutiny, a sloppy AI deployment is not a learning experience. It is a contractual problem.
So when MD James Walkinshaw decided to deploy AI Agents, he could not afford to get it wrong. And he didn't. Not because they moved fastest. Not because they bought the most tools. Because they asked the right question at the start.
Sixteen interviews before any vendor was contacted
Before any software was chosen, before any vendor was briefed, the leadership team laid out their entire creative process from start to finish and asked where time was being lost and where the work was suffering.
Then they interviewed sixteen team members. Individually. At every level of the business. Not a survey. An hour each. The question was not what AI tools would you like? It was what does your day actually look like, and what gets in the way of success?
That is the work that the 92% skip. And it is the work that determines whether what you build later actually gets used.
“We didn’t come with a preconception as to what the final output would be. We started with a blank sheet of paper rather than a specific area of the business or a piece of software. We really started looking at it from a holistic point of view.”
Out of those interviews came a set of guiding principles that have held from day one to now. Humans at the centre. One interface, not many. Scalable and flexible by team. Customisable by role.
The discipline of saying "should we"
The other discipline that mattered, and which James talks about often, is the discipline of focus. In agencies, people are constantly coming to leadership with ideas. Oh, we could do this. We could sell that to clients. James's response was consistent. Technically yes, but should we, and why?
“Everyone in this space is promising to deliver you the earth. You’ve always got to be focused on what’s it going to achieve, and therefore, should we proceed, before jumping in and losing where you’re trying to get to.”
That discipline is what protects an AI build from becoming a feature catalogue. The brief that finally went out to vendors was not a list of tools they wanted. It was a specific articulation of outcomes needed, constraints in play, and the experience they wanted for their team. And because the problem was articulated that precisely, the solution that came back was better than what they had imagined going in.
“James articulated the problem so well that we solved it in a way he didn’t expect. What he’s got now is a team where all the administrative tasks that used to be billable time are billable again. The same-sized team that can now service more clients, and even better.”
Notice the reframe. The question was not how much money can we save by cutting people? It was how much more can the same team deliver when the grind is removed? Those are different questions. They produce different answers. And only one of them produces a team that wants to use what you build.
This blog is part of a 4 part series. The second post in this edition covers what we actually built together, and why the most important thing was not an agent at all. It was the surface the agents lived behind.
Hear the full conversation between James Walkinshaw and Polly Barnfield on the FutureWeek podcast. Listen here.
Or, if you are wrestling with where to start your own AI deployment, book a call with the Maybe* team.
