The hardest question in AI transformation: “So what?”

Two years into HubSpot’s AI-first journey, adoption and fluency turned out to be the easy part. The real work, says its talent chief, lies in a “missing middle” most organisations never see. Ben Putterman, Vice-President of Learning and Talent Development, HubSpot, speaking at Reimagine ’26 in Singapore on 16 July 2026. Ben Putterman does not claim to have the answer – and he is suspicious of anyone who does. “When I go to conferences and any organisation tells me they’ve figured everything out, I tend to get a little sceptical,” HubSpot’s Vice-President of Learning and Talent Development told an audience in Singapore at the organisation’s Reimagine ’26 event, before sharing what has and has not worked in HubSpot’s two-year push to become an AI-first organisation. The uncomfortable question at the centre of that journey, he said, is one every leader will eventually face from their CEO, their board, or their employees: so what? “With all the activity, all the investment, all the talk that we hear about AI, the question is: what actually changed? What business results have you actually achieved?” HubSpot’s experience suggests the answer does not come from adoption alone. The organisation opened access to AI tools broadly in late 2024, backed by visible role modelling from CEO Yamini Rangan – who shared her own adoption journey, the tools she tried, and what worked or did not – and a deliberate push for experimentation, with employees told that failures would be as instructive as successes. The numbers looked healthy: 84% of employees said they felt comfortable using AI, and 90% were using it. Putterman was candid about what he would do differently. Opening the floodgates on tools “created a lack of clarity” about what employees should use for what, and the invitation to experiment came with too little direction on where. In 2025 – internally dubbed the year of fluency – organisational-wide learning days, hackathons and reverse mentoring followed, and AI fluency became a requirement in all hiring and promotion decisions. Even “fluency” needed sharpening: the term meant little at organisation level, HubSpot found, until it was defined function by function, because fluency for an engineer looks nothing like fluency for a salesperson. By year’s end, around half of employees reported using AI to automate or augment parts of their jobs – and, in a result Putterman did not predict, the organisation’s best hackathon came not from engineering but from its legal team. The work the organisation cannot see Much of that work, Putterman observed, lives nowhere on an organisational chart or process document. “It actually sits inside people’s heads” – the expertise and instincts that tell a salesperson when to pivot or which lead is worth pursuing. “AI cannot help you transform the work that it can’t see.” Surfacing that hidden work is the first step in the loop HubSpot now uses: see the work, redesign it with AI, measure, and scale. Seeing the work means leaders getting close to it – interviewing people, sitting beside them, and documenting how tasks are genuinely done rather than how the process flow says they are. Redesign then presents four options: automate a task; augment it by pairing an employee with an agent; stop doing it altogether; or deliberately keep it human where judgment, trust and relationships are at stake. Two cautions came with the framework. Teams should pick narrow, specific use cases rather than broad ambitions, and measurement cycles should run six to 12 weeks – anyone planning to check results in a year is “way off track.” The more profound lesson was human. Telling teams you want to understand exactly how their work gets done can easily read as a prelude to automation. HubSpot’s internal mantra: expose the work, not the people. “You make heroes out of those people [with knowledge in their heads], and you help them co-create the change,” Putterman said, adding that AI transformation “is as much, if not more, of a human transformation than it is a technology transformation.” Employees, he noted in conversation with HRM Asia afterwards, are not simply resistant or enthusiastic but both at once, feeling the opportunity and the fear of AI at the same time. When experience and title matter less That duality may be felt most acutely in hierarchical, tenure-based organisations, because of a cultural shift Putterman described as AI becoming “the great levelling tool.” The distinction between two years of experience and 20, he said, “is getting very, very blurry.” Junior employees at HubSpot are now coaching senior leaders – reverse mentoring that, in his experience, never previously worked. “I think some organisations will adapt well to that type of cultural change, and I think some will struggle with it tremendously.” The same recalibration applies to how HubSpot assesses its leaders. The single biggest focus is how they lead change and navigate ambiguity, alongside experimentation, clarity of priorities, and a willingness to get close to the work rather than operate above it. Putterman also offered a personal view that cuts against HR orthodoxy: the industry has the engagement equation backwards. Rather than engagement driving performance, “I think it’s performance that drives engagement. When employees are growing, they’re learning, they’re doing great work that makes a difference – that’s true engagement.” Leadership, he added, has drifted towards “a likeability scale versus actual leadership effectiveness.” If much of his message is about what must change, Putterman is equally clear about what has not. After three decades spanning Oracle, Tesla, and LinkedIn, his mantra remains “human first, managers second” – a recognition that every organisation “is just a collection of human beings.” And the first thing any organisation owes its employees, through every wave of technology: clarity – on what matters, on priorities, on performance, and on where their careers are heading.