OODA Loop – the new AI playbook

Product development speed is no longer dictated by the team’s capacity to ship code; it is defined by the speed of our strategic decision-making. We have all invested heavily in agile methodologies for years, organising our teams into two-week sprints, aiming for incremental, predictable delivery. That approach is now a bottleneck.

In the Sports & Live Broadcast sector, our products interact with Generative AI models that drift, learn, and iterate at a pace our existing structures simply cannot match. A two-week sprint cycle is an eternity when a model could be misinterpreting user intent or generating compliant risks every hour. The core challenge is not how fast we can write the next feature; it is how fast we can strategically pivot when the model’s behaviour changes. This is why I have focused on replacing our linear, predictable strategy cycle with the rapid, non-linear cadence of Boyd’s OODA loop.

Agile is a Slow OODA Loop

The problem with traditional Agile in an AI-driven environment is that it formalises the ‘Decide’ and ‘Act’ stages into rigid timeboxes, with ‘Observe’ and ‘Orient’ often relegated to retrospective meetings or quarterly planning. This is too slow for model drift and high-stakes performance environments.

Our job as Head of Product is to build systems that allow for constant observation, rapid orientation, and instant, compliant action. We must move beyond the framework and focus on strategic resilience.

I have found that the OODA loop provides the conceptual framework for this resilience. It is a four-stage process designed for a bias toward action and efficiency under extreme pressure.

Observe: The real-time performance dashboard

In an AI context, Observe means going beyond traditional product analytics. We must focus on model confidence scores, compliance triggers, and the immediate impact of generative outputs. This requires real-time monitoring infrastructure: a live dashboard that does not just tell us what the users did, but how the AI is behaving, including its ‘judgement’ calls and any deviations from baseline performance. The data must be organised for instantaneous strategic consumption, not leisurely analysis.

Orient: Rapid strategic re-alignment

This is the most critical stage for a Product Leader, and where Agile truly fails us. Orient is the process of filtering new observations through our existing context, culture, and strategic goals to form a hypothesis for action. With AI, this means rapidly assessing why the model drifted and immediately determining the new strategic constraints it must operate under. This is not a discussion for a planning meeting; it is a rapid-fire judgement call.

This need for constraint reminds me of the self-imposed musical discipline I undertook years ago; limiting my creative inputs to a single artist’s body of work actually forced a higher degree of focus and accelerated my productivity. In the same way, rapidly applying new, tight constraints to an AI model forces a clearer, safer path to a desired outcome, eliminating the latency of ambiguity.

Decide and Act: Automated, compliant deployment

Once a new strategic constraint has been Oriented, the Decide and Act phases must be near-instantaneous. The ‘Decide’ is codified into a system of automated, rule-based guardrails, which are then deployed instantly through automated pipelines in the ‘Act’ phase.

Our role is to design the systems and compliance checks that allow the model to automatically re-orient itself, or to trigger a safe, compliant rollback with minimal human intervention. We are building the rails, not driving the train. If a human must be in the loop for a decision, that decision must be made in minutes, not days.

Building Strategic Resilience, not Operational Efficiency

For the Head of Product in the AI era, success is not defined by perfectly executed sprints, but by the agility of our strategic mind-set. We must shift our focus from operational efficiency (making the two-week sprint slightly faster) to strategic resilience (slashing the time it takes to Observe, Orient, and Act). The OODA loop gives us the mental model to do this.

Our competitive advantage will be measured by the speed of our decisions, not the velocity of our code. The Head of Product who organises their team around this principle will win the AI race.