Commercial AI Transformation — Checkout.com

I lead commercial AI transformation at Checkout.com. It is the one role where I do both ends of the work at once: author the strategy and operating model, and build and ship the production agents that make it real. The remit is the commercial org — how a company that sells to enterprises restructures itself for the agentic era, and where agents are actually ready to take on real work.

The bet that shapes everything is redesign over augmentation. Bolting a chatbot onto an existing process gets you a faster version of a process that was already the wrong shape. The harder, higher-value move is to redesign the work around what agents can now do, and only then decide what a human still owns. That principle is the spine of the whole programme.

What I own

• The operating model. A workforce model and an executive plan for how the commercial organisation runs with agents in the loop — not a slide deck, the actual intake, prioritisation and governance that turns scattered AI ideas into a portfolio a commercial org can run and be accountable for.
• The portfolio. A 26+ use-case portfolio across the commercial funnel, each scored on whether an agent is genuinely ready to do the work or whether it is a demo that falls over in production.
• The builds. Production agents, not prototypes, with a human in the loop from day one. The one I own hands-on is a customer-voice and conversational-intelligence platform: it turns GTM conversations into a commercial knowledge graph, surfacing the signals and insights that sharpen how we sell.
• Adoption. The upskilling and change management so revenue teams use the tooling rather than receive it and quietly go back to the old way. A capable agent nobody adopts delivers nothing.

How I think about it

The scarce resource in a transformation is not model capability, it is trust and attention. So the work is pointed at the numbers that actually move — more qualified pipeline, faster onboarding, protected revenue, better operating leverage — and every agent ships with a human gate on the decisions that carry real commercial risk. Autonomy is earned per use case on evidence, not granted across the board because a demo looked good.

Where it is

I lead this at Checkout.com today, with the rollout still in its early stages. The patterns that prove out here are the same ones I use to reason about enterprise AI adoption generally: what to redesign, what to augment, and where an agent has genuinely earned the right to act on its own.