🎧 Listen to the full episode: Spotify | Apple Podcasts | YouTube
Direct answer: Governance failure is almost never a dramatic event. It is a gradual disconnect between what leaders believe is happening and what is actually happening on the ground, and organisations usually discover it after the point at which they could have shaped the outcome. Dhiren Katwa, who has spent two decades inside complex regulated environments including HS2, argues that the patterns preceding failure are sector agnostic, that the fastest diagnostic is how an organisation reacts to bad news, and that the true cost of excluding the people who carry the risk is not efficiency but legitimacy. For any organisation deploying AI, the implication is that governance is not the constraint on adoption. It is the precondition for it.
The Failure Nobody Hears
Ask most executives to describe governance failure and they will describe an event. A regulatory intervention, a public incident, a programme that collapses under scrutiny. The picture is dramatic, and it is almost entirely wrong.
Dhiren Katwa has spent 20 years inside some of the most complex regulated environments in the United Kingdom, including HS2, the largest infrastructure project in Europe. He is an award-winning PR practitioner, a diversity consultant and a stakeholder manager, and he has spoken at the United Nations in Geneva and at the inaugural World Public Relations Festival in Rome. His account of how governance actually fails is quieter and considerably more uncomfortable.
Governance failure is rarely a dramatic event. It is usually a gradual disconnect between what leaders believe is happening and what is actually happening on the ground. Dhiren Katwa
His working definition deserves to be written on the wall of every programme board. Governance failure is a situation where decision-makers no longer have an accurate understanding of operational reality, yet continue making decisions as though they do.
Everything a board would normally treat as the problem, the cost overruns, the stakeholder conflict, the regulatory intervention, the delays, the reputational damage, is on this reading a consequence of that single underlying failure. Which means most organisations spend their remediation effort on symptoms.
The Five Patterns
Dhiren names five recurring patterns that appear inside organisations before governance formally breaks.
Reporting becomes theatre. Decisions stop matching accountability. Stakeholder intelligence never reaches the top. Compliance replaces judgement. Bad news travels slowly.
What makes the list useful is that every item is observable without a review, an audit or a consultant. They are behaviours, not structures. And each is individually survivable, which is exactly why they accumulate unchallenged. No single instance of softened reporting looks like a governance failure. The aggregate is one.
In major change programmes governance is not primarily about committees or reporting structures. It is about ensuring uncomfortable truths can travel from the front line to the boardroom quickly enough for leaders to act on them. Dhiren Katwa
That reframe matters commercially. It moves governance from a compliance function, measured by the existence of frameworks, to an information function, measured by the speed and fidelity of what reaches the people making decisions. Those are very different things to be good at, and organisations routinely mistake the first for the second.
The Bad News Test
The most immediately applicable idea in the conversation is Dhiren's answer to how he distinguishes real governance from performative governance when he walks into an organisation. He does not begin with the governance framework, the committee structure or the assurance documentation. He watches what happens when someone delivers bad news.
In organisations with genuine governance, bad news moves quickly. People raise concerns early. Leaders ask questions rather than assign blame. Risks become visible before they become crises.
In performative governance the pattern inverts. People hesitate. Concerns are softened before they reach senior leaders. Language becomes euphemistic. And the messenger becomes the problem.
You can often tell within a few meetings whether people are speaking freely, or performing safety. Dhiren Katwa
The phrase performing safety is precise and worth sitting with. It describes an organisation where the appearance of psychological safety has been produced without the substance of it, where people have learned the vocabulary of open challenge and the practice of self-censorship at the same time. No framework detects this. It is only visible in behaviour.
Why This Is Sector Agnostic
Dhiren is careful about the limits of his own expertise, and that carefulness is itself instructive. His direct experience sits in infrastructure, environmental regulation, energy, local government and nationally significant programmes rather than in clinical settings. But his position is that the underlying patterns transfer, because governance failure is not domain-specific.
The variable that changes between sectors is scale, not mechanism. In infrastructure, governance failure affects multi-billion-pound programmes, national policy objectives and thousands of stakeholders. In a healthcare organisation it lands on patient safety, clinical quality, regulatory compliance, workforce pressure, financial sustainability and reputation in a local community. The consequences differ enormously. The path that produces them does not.
For any leader deploying technology into a regulated environment, that is the licence to learn from outside your own sector, and the warning that your sector's particular vocabulary is not protecting you from a general failure mode.
Alignment Happens Before Launch, Or It Does Not Happen
The most vivid passage in the episode concerns Dorothy, a 2,000-tonne tunnel boring machine deployed on HS2 to drive the first twin bore tunnel beneath Long Itchington Wood, a 400-year-old ancient woodland. Tunnel boring machines operate around the clock, ten metres underground, generating noise and disruption. Local scepticism was not irrational, it was proportionate.
What Dhiren did in response was not a communications campaign. Working from the HS2 visitor centre, he brought stakeholders, and the range mattered, MPs, ministers, schools, local businesses, to a viewing platform where they could see the scale and nature of the works for themselves. Then, rather than explaining engineering he did not personally command, he brought in the colleagues who genuinely understood the machine, and had them answer questions directly. Technical jargon was simplified rather than deployed as authority.
The outcome was not universal agreement, and he is honest about that. Not every mindset shifted. But a substantial number of sceptical stakeholders moved, because they could see why the work was happening and how it would eventually benefit them.
Be upfront, be honest, do not have any skeletons in the cupboard. Just say it as it is, and try to align key messaging with your different stakeholders. It works wonders. Dhiren Katwa
The transferable mechanism here is demystification. Something powerful and opaque was made legible to the people it affected, before it affected them. That is precisely the problem AI deployment now has in almost every sector, and almost nobody is solving it with a viewing platform.
The Cost of Excluding the Risk-Bearers
In most organisations, AI arrives through a narrow pathway. A vendor relationship, a budget holder, perhaps one or two internally enthusiastic champions, then implementation. The people who will actually carry the risk when something goes wrong, the practitioners, the front line, the regulator, the indemnity provider, sit outside that chain entirely.
Everything about that arrangement is genuinely efficient. Decisions happen faster. Procurement is simpler. Implementation is quicker. And if something does go wrong, the organisation can point to vendor assurances, technical documentation, contractual protections and regulatory ambiguity.
The cost is not primarily technical. The cost is legitimacy. Everything looks efficient, until the first failure. Dhiren Katwa
The observation that follows is the one with the sharpest commercial edge. In every regulated environment Dhiren has worked in, the organisations that recover most effectively from failure are rarely the ones with the best technology. They are the ones that had already built a coalition of stakeholders willing to navigate uncertainty together.
Where the risk-bearing stakeholders were absent from the formation of that coalition, an organisation may find it has implemented a system successfully and failed entirely to build the legitimacy required to sustain it. Legitimacy, unlike capability, cannot be procured after the fact.
Is Alignment Just Delay in Responsible Clothing?
I put the strongest counterargument to him directly. Stakeholder alignment is slow and expensive. It can function as a polite form of institutional risk aversion. And while organisations consult, the people who would have benefited from the technology are not benefiting. Is alignment a genuine precondition, or is it delay dressed up in responsible language?
His answer did not concede the framing. Alignment at the outset is what builds the trust that makes everything downstream faster, not slower. And he offered an observation about the technology industry that lands harder than a direct rebuttal would have.
Pointing to the calibre of speakers at London Tech Week, he noted that these are genuine experts in AI and technology, and then asked how many of them actually understand and align the stakeholders around the products they are building. The gap he is describing is not between fast and slow. It is between technical brilliance and the organisational competence required to make technical brilliance land.
Value For Whom
Asked whether the AI adoption conversation is asking who benefits and who might quietly be disadvantaged, Dhiren moved the discussion deliberately from governance to legitimacy. Every major technological change, he argues, eventually faces the same test. Who benefits, who bears the risk, and who gets left behind.
His observation is that organisations become fluent in functionality, efficiency and innovation long before they become comfortable discussing the distribution of benefit. The question is not whether AI will create value. It almost certainly will. The question is value for whom, and he is not convinced that is currently central to most adoption conversations.
The diagnostic he draws from the silence is worth quoting in full, because it is a test any leader can apply to their own organisation. If those questions are not being asked consistently, it may indicate that the technology is being treated primarily as a technical and commercial transformation rather than a social one.
Accountability questions rarely disappear. They simply arrive later, usually after deployment, when the opportunity to shape the outcome is much smaller. Dhiren Katwa
When a sector asks the distribution question openly, its governance is maturing. When the question is absent, the sector is still in the adoption phase of the conversation rather than the accountability phase. The accountability phase arrives regardless. The only variable is whether you are ready for it.
What Aviation Actually Got Right
Asked which sector has come closest to governing automation well, Dhiren names aviation, then immediately qualifies the claim in a way that makes it far more useful. Aviation did not get AI right. Aviation got governance around automation right, which is a different achievement and a considerably more transferable one.
The principle the industry institutionalised is that every incident is treated as system information rather than individual embarrassment. When something goes wrong, the first question is what the system allowed to happen, not who can be blamed. That single inversion creates learning, and learning creates resilience. Most regulated sectors, as he notes, struggle with it.
The lesson is not to become like aviation. The lesson is do not wait for your equivalent of an aviation accident before building the governance infrastructure. Dhiren Katwa
And the deeper point about what aviation actually built. The strongest feature of the sector is not its technology. It is that the industry created institutions capable of questioning, challenging and learning from technology. That is an organisational achievement, not a technical one, and it is available to any sector willing to build it before rather than after the event that forces it.
The Ten Year View
Dhiren's closing argument reframes what competitive advantage will look like once the technology itself has become ubiquitous. In ten years, he suggests, the most trusted organisations will not be the ones with the most advanced AI. They will be the ones that built governance, accountability, equity and trust into the technology before the technology became indispensable.
His emphasis throughout returns to oversight, and specifically oversight of stakeholder alignment and engagement. When he says stakeholders he means people, and he is insistent on that. Emotional intelligence and psychological safety are not soft additions to a technical programme. They are the mechanism by which a technical programme survives contact with an organisation.
At the end of the day, it is about people, and it is about taking people with you along on your journey. Dhiren Katwa
Asked whether any single country has got AI adoption right, he declines the premise, and the refusal is characteristic. We are a global village facing shared instability, and the more useful question is what each jurisdiction can learn from the others and where collaboration produces a better outcome than competition.
What This Means For Leaders, Founders and Investors
For leaders, the sequence is the strategy. Governance is not the brake on deployment, it is the infrastructure that makes deployment survivable. Run the bad news test on your own organisation, audit yourself against the five patterns, and identify the risk-bearing stakeholders currently absent from your decisions. All three are free and none require a consultant.
For founders and vendors, the uncomfortable read is that technical excellence is not sufficient and never was. The gap Dhiren identifies between brilliant technologists and stakeholder-competent organisations is a commercial vulnerability, because a product that cannot achieve legitimacy inside a client organisation will not be adopted regardless of its capability.
For investors, this points at a diligence question rarely asked. Weight a portfolio company's governance maturity and its stakeholder coalition alongside its technology, because the organisations that survive their first failure are not the ones with the best systems. They are the ones with enough legitimacy to be given a second chance.
The organisations that lead the next decade will not be the ones that adopted earliest or spent most. They will be the ones that built the capacity to know what is actually happening inside them, and to hear it in time to act.
Key Takeaways
Governance failure is a gradual disconnect, not an event. It occurs when decision-makers no longer understand operational reality but continue deciding as though they do.
Five patterns precede failure. Reporting becomes theatre, decisions stop matching accountability, stakeholder intelligence never reaches the top, compliance replaces judgement, and bad news travels slowly.
The bad news test is the fastest diagnostic available. Watch what happens to the person who raises a concern, and you will know whether people are speaking freely or performing safety.
The patterns are sector agnostic. Scale changes between industries, mechanism does not, which is why cross-sector learning is legitimate and sector-specific complacency is not.
Excluding risk-bearing stakeholders costs legitimacy, not efficiency. Organisations recover from failure through coalitions built beforehand, not through technology.
Ask value for whom before deployment. Accountability questions arrive regardless, and they arrive on worse terms once the opportunity to shape the outcome has narrowed.
Aviation's achievement was governance, not technology. Treat every incident as system information rather than individual embarrassment, and do not wait for your own accident to start.
This article draws on the TechDental conversation with Dhiren Katwa. Full episode on Apple Podcasts, Spotify, and YouTube.
About the Guest
Dhiren Katwa is an award-winning PR practitioner, diversity consultant and stakeholder manager with 20 years of experience inside complex regulated environments, including HS2, the largest infrastructure project in Europe. His work spans infrastructure, environmental regulation, energy, local government and nationally significant programmes. He has spoken at the United Nations in Geneva and at the inaugural World Public Relations Festival in Rome.
Connect with Dhiren: linkedin.com/in/dhiren-katwa
About TechDental
TechDental is a strategic intelligence platform for founders, executives, operators and investors shaping the future of dentistry. Through high-level analysis and systems-focused conversations, we explore how AI, governance frameworks and operating model design influence performance, scalability and enterprise value.
www.techdental.com | info@techdental.com