Conference floors are a peculiar kind of truth-telling. Strip away the keynotes, the polished demos, and the vendor positioning, and what remains — in the hallways, in the side conversations, in the questions organisations are actually asking — is a reasonably accurate picture of where an industry is at any given moment. I attended Gitex Asia in Singapore last week with that framing in mind, and what I observed is worth putting on record for anyone responsible for AI risk, governance, or operations in an enterprise context.
The headline stated plainly: AI deployment in the enterprise is accelerating faster than the governance infrastructure being built to support it. This is not a novel observation — it is the central concern of this newsletter — but GITEX Asia gave it a particular texture. The gap is no longer theoretical. Organisations are feeling it in production.
The Governance Gap Has Become Operational
The most significant pattern I observed across conversations at the show was not about technology selection. It was about what happens after deployment. Organisations that have moved beyond pilot and are running AI solutions at scale are now encountering something they did not fully anticipate: the need for ongoing monitoring and behavioral oversight of the systems they have put into production. The question being asked — with increasing urgency — is not “which AI tool should we use?” but “how do we know if it is still doing what we intended?”
This is precisely the accountability gap that governance frameworks have been slow to address. Deploying a model or an AI workflow is a tractable problem — there is no shortage of vendors willing to help. Knowing whether that model is drifting, behaving differently across edge cases, or producing outputs that no longer align with the risk tolerance it was configured against — that is a harder problem, and the market has not yet caught up with it.
Several organisations I spoke with had reached the same conclusion independently: the governance tooling available to them is either too generic, too compliance-oriented in a checkbox sense, or not instrumented for the operational realities of live AI systems. The gap between what procurement offers and what production requires is real and growing — and it is compounded by a capability problem that sits beneath it: the distance between what organisations say they want to build and the internal teams they actually have to oversee it.
The More Encouraging Signal
Not every organisation I encountered was in reactive mode. A meaningful cohort — what I would characterise as the more governance-mature enterprises on the floor — were approaching AI adoption with governance architecture in mind from the outset. They were asking different questions: not just what a system does, but how it behaves under variance, who is accountable when it doesn’t, and how accountability is operationalised rather than simply documented.
This is the right posture, and it deserves acknowledgement. It is easy to write about governance failures; it is worth being equally clear that some organisations are building in the right order — governance infrastructure before scale, not as a retrofit after something breaks.
The Deployment Wave Is Real and Accelerating
The breadth of AI activity at Gitex Asia was striking. Across virtually every vertical represented on the floor, organisations were actively evaluating and selecting AI solutions — not in an exploratory or experimental frame, but with the urgency of operational necessity. AI is not being treated as a technology of the future in these conversations. It is being treated as a capability gap that needs to be closed now.
The startup ecosystem reflected this clearly. There was a notable density of early-stage companies addressing highly specific operational use-cases through AI workflows and agents — not general-purpose platforms but pointed interventions in defined processes. This is a meaningful signal for governance leaders: the surface area of AI deployment in the enterprise is expanding not just through large platform vendors but through a growing long tail of specialised, often lightly governed solutions embedding themselves into operational workflows. Each of those integrations is a governance surface that needs to be accounted for.
The Ecosystem Is Fragmenting in Interesting Ways
One pattern that deserves specific attention is the behaviour of established software vendors and consulting firms at the show. A significant number were actively seeking to augment their existing offerings through partnerships with AI specialists — not to build AI capability internally, but to integrate it as a layer on top of what they already sell or deliver. This is a rational market response, but it has a governance implication that often goes unexamined: when AI capability is assembled through partnership and integration rather than built and owned, accountability becomes diffuse. Who is responsible for the behavior of an AI component that sits inside a product delivered by a consulting firm, built on a third-party model, and integrated into a client’s workflow? The answer, in most of the arrangements I observed being discussed, is not clear.
Hardware vendors presented a different kind of signal. AI is no longer confined to software and data pipelines — it is being embedded into physical infrastructure. Translation devices, object recognition systems, and action-enabled interfaces integrated into wearable hardware like glasses were visible across the floor. These are not edge cases. They represent a category of AI deployment where behavioral drift and governance accountability are genuinely difficult to instrument — the feedback loops are physical, the users are often frontline workers without governance context, and the output of a system error is not a wrong answer on a screen but a physical-world consequence.
The SuperNova Winner Is Worth Your Attention
Gitex Asia’s SuperNova competition — the conference’s recognition of the most compelling emerging technology — was won by Ailytics, a Singaporean company applying AI and video analytics to operational safety and productivity in heavy industries. Their approach is instructive in its simplicity: rather than requiring specialised sensor infrastructure, the platform converts standard CCTV into AI-powered monitoring tools — a design decision that dramatically lowers the deployment barrier and, by extension, the governance surface enterprises need to account for.
This is worth noting not as a product endorsement but as a directional signal. Heavy industry is not where most AI governance discourse is focused. The conversation tends to concentrate on financial services, healthcare, and knowledge work — sectors with well-developed regulatory environments and relatively legible AI outputs. But operational safety in industrial settings is a domain where AI behavioral failure has direct physical consequences — for workers, for assets, for communities adjacent to operations. The fact that this category of application is attracting serious capital and recognition should prompt governance leaders in asset-heavy industries to treat the question of AI behavioral oversight as a first-order operational risk, not a technology compliance matter.
What This Adds Up To
Gitex Asia was not a governance conference — but the signals visible beneath the surface activity tell a consistent story for those paying attention. The deployment wave is accelerating faster than the governance infrastructure being built to contain it, the accountability perimeter is larger and more complex than most risk frameworks currently reflect, and organisations that have already deployed are beginning to feel this acutely. None of this was a surprise. But it is useful, occasionally, to see it confirmed on the floor.