EP27: Build, Orchestrate, Scale: Agentic AI for Regulated Enterprises with Futuria
From Emerging Technology to Enterprise AI
Rob’s career has consistently involved applying emerging technologies at scale. His experience spans global businesses, consulting, digital leadership and operational roles, giving him a perspective that combines technology with the practical realities of running large organisations.
He has worked as a consulting managing partner, chief operating officer and chief digital officer, meaning he has encountered AI from both sides: as someone providing technology to organisations and as a leader responsible for deciding whether new technologies are robust, financially viable and operationally useful.
This became the foundation for Futuria. Rob and his co-founders, with backgrounds spanning consulting, intelligence and defence, identified an opportunity to apply generative AI within environments where security, confidentiality, personal data and intellectual property are critical.
The question was not simply how powerful AI models could become, but how organisations could safely harness that power.
Moving Beyond the AI Proof of Concept
One of the biggest challenges facing businesses is the gap between experimentation and operational deployment.
Rob believes organisations are right to experiment with AI. Businesses need to understand the emerging possibilities, identify the skills they will require and develop an understanding of how AI could change their operations.
The problem comes when experimentation never develops into something operational.
Rob draws a distinction between creating a proof of concept and building something that can operate reliably at enterprise scale. The latter requires organisations to consider governance, existing technology infrastructure, data, controls and the practical realities of deployment.
This is particularly important as AI becomes increasingly associated with agentic systems. While the technology can develop rapidly, enterprise organisations cannot simply adopt the newest capability without understanding how it fits within their existing environment.
Agentic AI Requires a Different Way of Working
For Rob, agentic AI represents a shift away from thinking about AI as another conventional software system.
Traditional systems are generally designed, built and deployed according to a defined specification. Agentic systems can instead be introduced with a particular capability and then continuously improved through interaction, additional data, tools, connectivity and training.
Futuria's approach is to configure multi-agent teams that can perform complex work within secure enterprise environments. These teams can combine different agents with human involvement, allowing organisations to determine where autonomous decision-making is appropriate and where human oversight is required.
The result is less about replacing an existing system and more about introducing a new capability into an organisation and developing it over time.
Building the Right “Harness” Around AI
For regulated sectors such as insurance, banking, defence and national security, the challenge is not simply whether AI models are capable enough.
Organisations need to understand the regulatory environment, the relevant data and the controls required to operate safely. Rob argues that organisations do not necessarily need to make every piece of their data perfect before experimenting. Instead, they should identify the specific datasets required for a particular use case and ensure that those are fit for purpose.
The wider architecture also matters.
AI needs to operate within the organisation's existing technology, data and governance environment. This means creating what Rob describes as a kind of harness around the models: the structures that allow organisations to use increasingly powerful AI while maintaining appropriate levels of security and control.
For highly regulated businesses, this can create a route towards experimentation without abandoning the risk management that the organisation depends on.
The Leadership Mindset for AI Transformation
Curiosity is one of the most important qualities Rob identifies in leaders working through AI transformation. Leaders need a willingness to experiment and understand what is newly possible.
But he also argues that leaders need to be willing to temporarily step outside established assumptions.
Organisations have spent decades developing particular approaches to data, technology and workflows. AI may challenge some of those assumptions. Rather than automatically reproducing existing structures with AI layered on top, leaders need to ask whether entirely different approaches are now possible.
This could ultimately change how organisations interact with data itself. Instead of relying exclusively on traditional structures such as data warehouses and data lakes, agentic systems could potentially create dynamic connections between disparate sources, accessing information when it is needed.
The important point is not that existing systems should immediately disappear, but that leaders should remain open to alternatives that were previously difficult or impossible.
Augmenting People, Not Simply Replacing Them
A recurring theme throughout the conversation is the importance of using AI to increase human capability.
Rob recalls one customer explaining their motivation for adopting the technology simply: “I wanted to make my people incredible.”
That idea provides an alternative to the common narrative that AI adoption is primarily about reducing headcount. Instead, organisations can use AI to give employees new capabilities and allow them to perform work more effectively.
At the same time, Rob recognises that AI will change the composition of the workforce. Organisations are already struggling to recruit certain capabilities, while demographic changes will place additional pressure on the workforce of the future.
AI therefore becomes part of a much broader question: what should an organisation's workforce look like in ten years, and what skills will it need?
The Evolving Role of the Chief AI Officer
The emergence of the Chief AI Officer raises questions about where responsibility for AI should sit within an organisation.
Rob does not believe there is one universal answer. Depending on the organisation, AI leadership could sit with a Chief AI Officer, Chief Data Officer, COO, CIO or another senior leader.
What matters is that someone has sufficient expertise to understand the rapidly changing AI landscape and help the wider organisation navigate it.
That role should go beyond evangelising technology. A strong AI leader needs to understand the depth of the technology, challenge poor ideas, identify new possibilities and help drive transformation.
Most importantly, Rob argues that AI leadership needs to be embedded within the organisation rather than positioned on the sidelines. AI transformation at scale requires someone capable of influencing budgets, governance, strategy and organisational change.
Designing High-Performing Human-Agent Teams
As AI agents become part of the workforce, organisations will need to reconsider how teams themselves are structured.
Building an individual agent is becoming increasingly accessible. The more difficult challenge is getting multiple agents to collaborate effectively and produce useful work.
Harder still may be encouraging human employees to engage positively with those systems.
Rob highlights an interesting opportunity around organisational knowledge. Many experienced employees will retire over the coming years, taking decades of expertise with them. Agentic systems could potentially capture and structure some of that knowledge through knowledge graphs and agent teams, allowing organisations to retain and reuse expertise.
But technology alone cannot solve the problem.
Organisations will need to think about incentives, motivation, ownership and commercial questions around knowledge. They will also need to work out what high-performing teams look like when some members are human and others are AI agents.
Humans in the Loop — and the Question of Trust
Human involvement remains an important part of Rob's approach to agentic AI.
Not every decision needs to be checked manually. One of the reasons organisations deploy agents at scale is precisely because they can perform work faster than humans.
Instead, humans can be positioned at strategic points within workflows, providing oversight and confidence that systems are continuing to operate as intended.
This creates a more nuanced understanding of the “human in the loop”. Rather than asking people to review every individual action, organisations can determine where human judgement adds the most value and where autonomous processes can operate safely.
For highly regulated organisations, explainability also remains important. Leaders need ways to understand what systems are doing and intervene when necessary.
AI's Ability to Unlock the Previously Impossible
Perhaps the most exciting part of agentic AI, according to Rob, is its ability to make previously impractical tasks achievable.
He describes a recent example involving a complex multi-agent system in the US. Initially, the team believed its output would need to be limited to formats such as Word or Excel because producing a sophisticated PowerPoint presentation automatically was considered too difficult.
Within two weeks, the system was producing a complex presentation incorporating charts, text, data, analysis and images.
Another example involved highly unstructured data that initially appeared difficult to use. The team was able to ingest it and build an agent team around it within a day.
For Rob, these examples demonstrate the speed at which the boundaries of what is technically possible are moving.
The challenge for organisations is therefore not simply implementing today's technology, but continuously identifying what has become possible since yesterday.
The UK Opportunity: Application Rather Than Foundation Models
Rob is cautious about positioning the UK as a direct competitor to the US or China in the race to build increasingly powerful foundation models.
Instead, he sees a significant opportunity in the application layer.
The UK has deep expertise across professional services, financial services, government and other complex sectors. These environments provide an opportunity to develop sophisticated applications of AI and agentic technology.
Importantly, Rob argues that no country has yet fully “cracked” agentic AI. The technology is still relatively young in terms of practical enterprise deployment.
The opportunity for the UK is therefore to move quickly, build practical expertise and scale successful applications rather than waiting for a definitive global model of how agentic AI should work.
Preparing the Next Generation of the Workforce
The changing nature of work also creates an important opportunity for younger workers.
Rob believes people entering the workforce now can bring fresh knowledge and fewer assumptions about how work has traditionally been done. They may be particularly well positioned to experiment with emerging technologies.
However, there is a tension: these same early-career roles may be among those most affected by automation.
Tasks historically performed by junior consultants, for example, could increasingly be undertaken by AI agents.
This means organisations and policymakers need to think about how to create meaningful opportunities for people entering the workforce while still benefiting from their technological fluency and willingness to experiment.
The future workforce will need both the accumulated experience of established professionals and the fresh perspective of people entering an increasingly AI-enabled world.
From Experimentation to AI at Scale
The next stage of AI adoption will require organisations to move beyond simply asking what AI can do.
Rob's closing challenge is to ask what needs to be true for an organisation to adopt AI and agents at scale.
That means thinking about safety and trust, governance and control, quality and performance, data, organisational capability and the structure of high-performing human-agent teams.
The technology will continue to evolve rapidly. The organisations that benefit most may therefore not be those that simply adopt the newest model, but those capable of continuously adapting their people, processes and governance around what the technology makes possible.
For Rob, the central challenge is turning AI's extraordinary rate of innovation into something that can operate safely, effectively and at scale.