Become a TechRadar Insider
Across cloud, SaaS, and modern AI tools, teams can now move from idea to working prototype in days.
What once required long procurement cycles, heavy engineering effort, and multiple layers of approval can now be assembled with off-the-shelf services and a handful of integrations.
That speed is real, and it is welcome. But it has changed the underlying economics of decision-making in a way most governance models have not caught up with.
Founder & CEO of Emergn.
When the cost of starting drops, the volume of starts rises. When the volume rises, the cost of being wrong compounds, because misjudged initiatives no longer stay contained inside a pilot.
They spread into systems, workflows, and roadmaps before anyone has a clear read on whether they should exist at all.
The bottleneck has shifted. It is no longer how fast you can build. It is how clearly you can decide what deserves to keep being built.
Governance frameworks
Most organisations are still running governance frameworks designed for a slower pace, when far fewer ideas move forward at once. Those frameworks assume the hard problem is execution: once something is approved, the job is to deliver it. In an AI-rich environment, that assumption is backward. The harder problem is continuous selection: repeatedly deciding whether an initiative still deserves resources as the context around it changes.
This is where the idea of a kill engine becomes useful. Not as a slogan, but as a practical discipline embedded into how an organisation runs its technology portfolio. A kill engine is a deliberate system that makes it normal, expected, and evidence-based to stop initiatives early when they no longer show sufficient value. It treats every active piece of work as a capital allocation decision rather than a permanent commitment. Continuation must be earned, not assumed.
Vague strategic intent
In practice, this means a few things. Initiatives are funded against pre-agreed value hypotheses, not vague strategic intent. They are reviewed on a monthly cadence, sometimes more often, against evidence rather than enthusiasm. Stopping criteria are written down at the start, when judgment is clearest, rather than improvised at the end, when emotion and sunk cost are loudest. And cancellation is rewarded as a positive signal, not absorbed as a quiet failure.
This sounds almost too basic to matter, but it directly challenges a deeply ingrained behavior in enterprise environments: once something starts, it tends to continue by default unless there is a strong reason to stop it. A kill engine reverses that default.
The asymmetry that has always existed, where stopping feels like a negative signal even when it is the most rational option available, is precisely where cost and complexity build up. Removing that asymmetry is the single highest-leverage governance change most organizations could make this year.
Introducing discipline
When this discipline is introduced, behavior shifts quickly. Teams become more explicit about their assumptions from the outset, because they know those assumptions will be tested. Leaders become more comfortable ending work that no longer shows meaningful progress, because the system, not the individual, is making the call. And the slow accumulation of low-value initiatives, the quiet drain on attention, talent, and budget that every large organisation knows but few measure, begins to reverse.
AI will intensify the pressure to get this right. As organizations adopt generative systems, agentic tooling, and embedded intelligence across their stack, the volume of plausible initiatives will continue to rise. More capability produces more ideas. More ideas produce more partially-validated commitments competing for the same finite leadership attention. Without a structured mechanism to remove the weakest ones, complexity compounds faster than clarity. The organisation stays busy, but focus weakens.
The organizations that navigate the next phase of AI well will not be the ones that build the most. They will be the ones with the discipline to keep deciding, in public and on cadence, what still deserves to be built. AI does not just change how quickly organizations can move. It changes how urgently they need to choose.
In that environment, the ability to stop weak ideas early is not a defensive behavior. It is a core operating discipline, and increasingly, a competitive one.
We feature the best IT automation software.
This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.
The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit