Preeti Verma on the End of Headcount Planning, the AI Fragmentation Tax and Formula One

Every so often a guest reframes something so cleanly that you find yourself rethinking a question you thought was settled. Preeti Verma did that with headcount.

Preeti Verma is Strategic Initiatives & Transformation Lead at Atlassian, based in Australia. Her twenty-plus year career spans Johnson & Johnson, General Electric including an international posting to Connecticut, and Accenture before joining Atlassian —where she now sits at the intersection of business teams, people strategy and AI transformation. Ram caught up with her on a Friday evening in Australia for a conversation about why workforce planning is breaking, what most companies get wrong after buying AI tools, and how a Formula One team ended up proving Atlassian's entire thesis.

What stood out

Headcount is no longer the right unit of measure.

This was the heart of the episode. For as long as anyone has been doing workforce planning, capacity has meant human capacity: how many people, at what cost, to deliver what output. Preeti argues that unit is breaking.

"One person with the right AI leverage can now do three, four, ten times what humans could do earlier. So the question shifts from how much headcount do we need, to what output do we need — and what's the best combination of human plus AI plus workflow."

‍

The AI fragmentation tax.

Atlassian's State of Teams research surveyed 12,000 knowledge workers and 170+ Fortune 1000 executives. The headline finding: 89% of executives say AI has increased the speed of work, but only 6% can point to organisation-wide ROI. Preeti's diagnosis of that gap is sharp — it is not a technology problem. It is a coordination problem, a workflow problem, and a culture problem.

Individual productivity gains go nowhere without the value chain.

Preeti made this concrete with an example any organisation will recognise. Give your engineers AI tools and they code faster. But they still depend on designers upstream to hand over the work, and on marketing and sales downstream to take it to market. Speed up one link and the output simply piles up waiting for the next.

"There's no point in individual productivity if the whole squad isn't coordinated."

‍

Ram framed it as a relay: the baton still has to get passed. Productivity unlocked at one station in the chain, without the rest keeping pace, produces nothing an organisation can actually bank.

Stop measuring individual AI adoption.

Her most emphatic point, and one aimed squarely at HR leaders. Atlassian went through three rounds of measurement strategy before landing here. Leaderboards tracking who burns the most tokens drive exactly the wrong behaviours — people running processes in the background just to climb the rankings. The right unit of measurement is the team: how fast are they shipping, how many handoffs have been eliminated, how much duplicated work has gone.

Overlaying AI on old workflows is patchwork.

Ram offered a useful analogy here: you cannot pull the engine out of a petrol car, drop in a battery and call it an EV. You design for an EV from the start. Preeti's version is the same principle applied to work — redesign the end-to-end workflow with AI as a teammate from the ground up, rather than bolting it onto processes built for a human-only paradigm.

Give people permission to stop doing things.

The third of her three recommendations, and the one that requires leadership air cover. Most people are still doing their old job in the old workflow with AI layered on top. Preeti's point: people need explicit permission to say I am not doing this task anymore because AI does it better — and that permission has to come from the top.

Formula One as proof of concept.

Atlassian became title and technology partner of the Atlassian Williams Formula One team in early 2025. Co-founder Mike Cannon-Brookes was clear from day one that this was never about a logo on a car. Behind every driver sits a team of roughly 800 people across engineering, aerodynamics, manufacturing, race strategy, logistics, HR and marketing, all coordinating in real time. Preeti described it as the most extreme team sport in the world, dressed up as an individual sport.

The wind tunnel example brought it home. Testing a Formula One car generates enormous volumes of raw data that traditionally only three or four specialist engineers at Williams could interpret. Everyone else waited. Williams trained a Rover agent on that data, and now any engineer can ask a plain-English question and get a contextual, actionable answer in seconds. The bottleneck disappeared.

One thing I'll keep thinking about

Preeti closed with a version of the future of work worth holding onto. AI is not there to replace the coffee walk with a colleague. It is there to clear the grunt work that was stopping that walk from happening in the first place — the ten million small things that kept you at your desk when you should have been out bouncing ideas off a teammate.

That is a considerably more optimistic framing than most of what is currently in circulation. It also happens to be the one that matches what this podcast has been about from the start.

Enjoyed this episode? Subscribe to Fika Friday at the Office and share it with someone still building next year's plan around headcount.