Andrew Grill delivered the guest keynote at Adobe’s Closing the AI Productivity Gap event at Adobe Old Street in London on 6 October 2026. Speaking to Adobe’s enterprise customers from IT, procurement, marketing, legal, HR and finance, he tackled the question every organisation is now asking: if AI is everywhere, why is it still so hard to get value from it? His argument was that AI will not replace jobs, it will slice tasks out of them, and the organisations that pull ahead will be the ones that decide which tasks to slice first.
Focus on what you love, automate the rest
Andrew asked the room to picture their job not as one thing but as a hundred tasks. The tasks worth handing to AI pass two tests: they are repeatable, and they are data-rich. Expenses are the classic example. Drawing on his work with SAP Concur, he described the shift from expense management to exception management, where software queries only the £50 coffee that needs explaining and lets everything else through. In sales, he played a clip of Dell’s John Roese explaining that 40% of the sales force’s time went on preparing to meet customers, and that winning time back for customer conversations would be a material advantage. People train for years to use their judgement, Andrew said, not to chase receipts or rebuild the same proposal at 11pm.
Find your Wally
The word picture of the morning was a Where’s Wally beach scene. Andrew gave an AI tool the picture with no context and asked it to circle Wally, and it did. That crowded beach, he told the room, is your organisation, and somewhere in the noise is the one fact you need for tomorrow’s report. He backed it with real stories. A 17,000-cell SWOT analysis that had taken a manager 10 days to review was worked through department by department in a single morning, and a sceptical fifth-generation CEO left the first coffee break with eight pages of notes and a decision to proceed. A law firm used a specialist AI tool to search 150,000 documents, many of them in Russian, and found the conversation that won the case.
Not a skills problem, a permission problem
Andrew drew on Adobe’s 2026 AI and Digital Trends research throughout. 57% of organisations agree AI is changing work faster than employees can adapt, and 47% say executives and practitioners are only partly aligned on AI strategy. 71% blame a skills gap, yet only 45% have an AI training programme good enough to use, and 52% cannot demonstrate a return on AI with their existing metrics. His route to ROI is practical: break a process into its parts, cost the old way, then cost the AI-assisted way. He also urged leaders to model curiosity from the top, because teams copy what they see rewarded and avoid what they see ignored.
In the Q&A, Andrew advised asking AI to justify its answers and cite its sources to catch hallucinations, and warned that voice cloning now makes a family password and a team password essential, citing the deepfake call that cost the engineering firm Arup $25 million.
What this means for leaders
- Pick one task, not a project. Start with something repeatable and data-rich that nobody will miss doing by hand.
- Say the permission sentence out loud. Tell your team explicitly that experimenting with AI is part of the job.
- Check what you are already paying for. Ask every software provider whether AI features exist in your current licence and switch them on.
- Pick one metric and watch it for 90 days. Speed, cost or query volume. If the needle moves, take the evidence to the board.
Attendees received free access to Andrew’s Curious Executive course and his Curiosity Quotient self-assessment, which takes about three minutes at curious.courses/cq.
