Your AI Strategy Doesn't Need a Task Force. It Needs a Home.
Most companies respond to the AI moment by forming a task force. A cross-functional group meets biweekly, shares updates, and produces a slide deck that no one acts on. Six months later, someone asks what happened to the AI strategy.
The problem isn't effort or intent. It's that task forces are designed to explore, not to execute. They have no budget authority, no headcount, and no product owner. When an AI initiative competes with a revenue-critical feature for engineering time, the task force loses every time.
The alternative: give AI a home
The organizations making real progress on AI share a common pattern. They've assigned a senior leader whose explicit job is to own AI direction — not as a side project, but as their primary mandate. This person has:
- Budget authority to make build-vs-buy decisions without committee approval
- A seat at the product table, so AI capabilities shape the roadmap rather than chasing it
- Technical credibility to evaluate what's real versus what's demo-ware
- Executive sponsorship to clear organizational blockers
This isn't a Chief AI Officer title for the sake of having one. It's a recognition that AI strategy requires the same sustained attention we give to security, compliance, or platform architecture.
Build, buy, or orchestrate
At Recurly, I introduced a three-option framework for evaluating every AI investment: build it, buy it, or orchestrate it. Most companies only think in terms of build versus buy. The third option — orchestrating best-of-breed AI services behind your own product experience — is consistently underinvested.
Build when you have a genuine data moat. If your proprietary data creates a model or capability that competitors can't replicate, building is justified despite the cost.
Buy when the capability is commodity. If five vendors offer roughly equivalent solutions and the technology isn't a differentiator, don't build it.
Orchestrate when the value is in the integration. Many of the best AI-powered features aren't about having the best model — they're about connecting a good model to your specific domain data and user workflows.
Making it stick
The hardest part isn't picking the right framework. It's maintaining conviction when the first few AI experiments don't produce immediate ROI. AI strategy is infrastructure work — it compounds over time but rarely pays off in the first quarter.
Give AI a home. Staff it with someone who's shipped under real constraints. And give them the runway to build something that lasts.