AI adoption
An AI squad, not a committee
AI is not a technical problem. Don't give it to IT to own.
AI is not a technical problem. Don’t give it to IT to own.
That’s what a senior member of an enterprise organisation told me recently, and he’s quite right. IT should own the AI infrastructure and the harness: including the boundaries, the tools, containment, scaling and security. But individual teams should own the workflows themselves. It’s collaborative ownership - not a single group or a single team.
AI Labs and champions
A lot of organisations are creating AI Labs - finding AI championsGitHub’s playbook builds a volunteer network of champions. This organisation made it their entire job from across the business. Inside these labs are squads, not committees. They’re action groups. They own the outcomes. This organisation removed their previous responsibilities, so AI becomes their entire focus.
The squad is where experiments and pilots can land, and (critically) expand.
Replace old sign-off
These AI Labs need to replace old sign-off processes with new, AI-first ones. To be successful, they often need an internal C-suite champion: someone who provides ‘air cover’Adobe’s 2026 report: executive misunderstanding of AI is the top cause of misalignment (61%) for the new processes.
What IT should build
Let IT build your AI system architecture, providing a safe and scalable environment for your organisation. But let your AI Lab innovate and experiment.