“The Robots Are Coming, But Where Will They Sit?”   

Leading through AI uncertainty & the search for real agentic leadership 

AI and agentic transformation is set to become the biggest paradigm shift in technology since the launch of the internet. The ability for AI to not just automate tasks (I mean, that’s just fancy RPA right?), but to allow companies to reimagine how they interact with, view and understand their data, their customers, suppliers, and pretty much all other involved parties is set to revolutionise pretty much every industry. It’s real blue sky thinking space and sets the mind ablaze with opportunity.  

The leadership dilemma

However, leaders of many of today’s businesses are still struggling to develop a coherent strategy around use-cases, deployment strategies and what “end state” should look like, and how to communicate it.  

To compound this, some companies are now set with the daunting task of running transformation agendas: 

• Concurrently modernising multiple generations of obsolete technology to bring their estate into appropriate shape for meaningful deployment of AI;  
• Reimagining their business model;  
• Providing direction, stability and coherent messaging for their workforce. “Yes, the robots are coming, but we haven’t worked out where they’re sitting yet..” 

Echoes of Blockchain?

It bears some passing resemblance to Blockchain in the late 2010s, where there was almost 100% adoption of Blockchain in the management strategy packs of most financial services and fintech companies, but they were struggling to find and deploy actual use cases through to everyday company usage.  

Admittedly, the use of Agents is (at least on the surface) an awful lot more accessible and tactile than immutable, hyper secure distributed ledger technology. Most people should be able to find something positive in the use of AI within their workflows. 

Outrunning the bear

We are receiving consistent messaging from our clients that everyone sees the need to change, from both an offensive and defensive strategy, but often don’t know where to start.  

Everyone is worried about AI Native companies being the grizzly that crashes out of the bushes to ruin a nice walk. But fundamentally, we believe for many companies with solid propositions, reasonable market share, user base and proprietary data and/or IP, we don’t think you need to outrun the bear… you just need to outrun the slowest hikers. 

The missing candidate profile

Our job in Executive search, isn’t to provide you with the answers, but to help you find the people who can find the answers with you, whether they are advisors or full-time operators.  

The challenge with this, as with all new transformational technologies, is that we can’t find someone with “10 years of Agentic AI transformation experience” (not heard this one yet, but I’d bet cash money it’s coming).  

So, to try to help our clients solve this problem, the brief must change to focus on a new take on understanding the core components of technical and transformational leadership in the AI & Agentic era.  

We believe there to be 6 key pillars to review when selecting leaders, to avoid “innovation theatre” and a post Blockchain hype boom-esque “trough of disillusionment”:  

Foundational Skills

Digital & technology transformation experience

• A track record of actively delivering large-scale transformation, not just sponsoring it from a safe distance 
• Deep, hands-on understanding of modernising legacy and multi-generation technology estates 
• Ability to move at pace across multiple, concurrent and disparate workstreams  

Change management experience

• The Kotter Lineage: This is fundamentally a people problem wearing a technology costume. Over 30 years ago, John Kotter (A Force for Change, 1990) drew the distinction between remote management (administering from a distance) and real leadership (carrying change yourself through continuous direct action). 
• Delivered, not sponsored: We test for leaders who actively lead change from the front, rather than “sponsoring” it from the safety of a steering committee. 
• Cultural execution: The soft skills to take a skeptical workforce through deep cultural change, preparing teams for adoption rather than just celebrating a technical rollout. 

Stakeholder management & communication

(tied to the above, but also a specific and standalone point) 

• Able to articulate strategy from Board to frontline workers, and to manage up and down 
• Skilled at turning a vague mandate (“go and do something with AI”) into a narrative people can actually follow, repeat, and buy into 
• Providing a clearly articulated mechanism to deliver the narrative. 

Key AI skills

Innovation at scale mindset

• A demonstrable track record of driving innovation inside an organisation, not just alongside one 
• Not necessarily AI-specific but shows the ability to prototype, experiment, and get an idea through to real adoption, rather than letting it die quietly in a pilot 

Applied AI fluency

• Evidence of having genuinely used modern AI tools in anger (not just “a computer vision project 5 years ago”), ideally at some scale, but at minimum with a properly solid grasp of how they actually work, where they still break and where the value can be sourced 
• Pragmatic AI Governance: An understanding of risk, data privacy, and security that enables safe adoption rather than using compliance as an excuse to stall momentum. 

What would you do differently now?

• The self-awareness to look back at transformation programmes they’ve run before including the ones that didn’t fully land and say plainly what they’d do differently, knowing what agentic tools can and can’t yet reliably do 
• Enough intellectual honesty and understanding to know what AI still can’t do well  

Lessons from History

Whilst this current wave of technology is new, we believe there are lessons to be learned if we look back to the dawn of the internet (and possibly beyond).   

Dot.com boom

• The original hype cycle. Companies bolted “.com” onto the name and doubled the valuation overnight, promising a frictionless, borderless future for absolutely everything.  
• Most of the specific promises didn’t land the way anyone predicted at the time, but the underlying shift was real, it just took the better part of two decades and a spectacular crash to actually arrive. 

Cloud transformation

• The first wave of cloud migration had exactly this problem. Job specs asked for five years of AWS experience when AWS had barely existed for two. 
• Companies made the jump anyway, by hiring for infrastructure judgement, architecture experience and the ability to learn fast, rather than waiting for a “cloud-native” candidate pool that didn’t exist yet. 
• We think agentic AI is the same trade-off, just compressed into a much shorter window. 

Fintech

• A few years ago, I ran a piece of research tracking where fintech leadership teams had actually come from. The number that stuck with me: roughly 80% of the leaders we tracked had been hired from outside the specific fintech vertical they went on to lead.  
• The takeaway wasn’t that fintech expertise didn’t matter; it was that the underlying leadership skillset mattered more than sector-specific years.  
• We think the same will hold here. With a current lack of candidates with direct experience in delivering AI & Agentic transformation in companies, hiring teams and ExCos will need to find work arounds to fill these requirements.  

Innovation departments

• Every large enterprise built one of these in the last decade: a shiny “Innovation Lab,” usually with beanbags, a 3D printer nobody used and a mandate to “disrupt” the core business from a converted meeting room in a separate building.  
• Most of them produced excellent innovation theatre and vanishingly few things that made it back into the P&L. The risk with agentic AI is the same one, just with better demos delivered faster. 

Blockchain

• To wrap up the blockchain comparison, most deployments were technology looking for a problem, not a problem looking for a technology. Gartner’s own research found blockchain “interest” tracked the price of bitcoin far more closely than it tracked any actual business case. 
• Outside of crypto trading itself, genuinely proven at-scale production use cases stayed rare for the better part of a decade. 
• I don’t think AI has quite the same issue regarding use cases, but I will be curious to see how the corporate interest in AI is affected as hype waxes and wanes. 

Conclusion

The internet eventually delivered on its promises, but it took twenty years, a market correction and many failed bets on hype over people. Agentic AI won’t be as patient. Whatever separated the winners from the losers last time is about to play out in fast forward. 

The businesses standing when the dust settles won’t be those that deployed the most agents the fastest. They will be the ones that put the right human judgement in the room before pulling the trigger: leaders who know how to take a workforce with them, evaluate what the technology can actually deliver, and execute against commercial outcomes. 

Everything else is just a much quicker way of finding out who the slowest hiker is. 

Footnote from a Sci-fi fan 

As an insufferable nerd and Sci-fi fan, I am fascinated by authors like Alastair Reynolds (Revelation Space) and Hannu Rajaniemi (The Fractal Prince), who take current scientific concepts and extrapolate them far into the distant future and weave them into their stories. I was fascinated to hear (years ago) how Nasa scientists were taking huge inspiration from things like Star Trek and Kim Stanley Robinson books (the Mars trilogy).  

I also re-read the legendary Isaac Asimov novel Foundation and Empire in the early 2010s, and was immediately struck by the almost exact match of the description of Psychohistory and how it mapped EXACTLY at that time to how people were describing Big Data and Predictive Analytics (remember those?). 

The old Alan Kay adage holds: “The best way to predict the future is to create it.” Sci-fi writers haven’t just predicted where we’re going, they’ve started to give us the blueprint to go out and build it. 

Personally, I will continue using sci-fi for inspiration in how we can use AI, and I think it will be time well spent!