EP10: Building AI Foundations: Fixing Flow, Structure & Strategy with Lumo

"No AI for the Sake of AI"

Lumo's guiding philosophy is refreshingly blunt for an AI consultancy: don't adopt AI without first being crystal clear on why. Joe explains that too many organisations pursue AI from a top-down, competitor-driven mindset, without a clear problem to solve — wasting time, money, and effort in the process. Lumo's framework instead centres on five key pillars: a clear purpose, data quality and accessibility, technology architecture readiness, the operating model and flow of work through an organisation, and — the biggest piece — the people and workforce who ultimately have to trust and adopt any new capability.

Learning the Hard Way: Why AI Isn't Plug-and-Play

Joe and Krantik's own journey into AI consultancy began with a near-acquisition of an AI services firm at their previous company. Initially convinced AI would be a natural, easy addition to their existing services, it took them roughly two years to realise otherwise: AI requires deep integration into an organisation's operating model, data flows, and structures — not a simple bolt-on. That hard-won lesson — trying to implement AI "with all the constraints of an organisation" rather than removing them — became the founding insight behind Lumo itself.

The Building Analogy: Why Siloed AI Projects Stall

One of the most memorable frameworks from the conversation is Krantik's "building" analogy. Large organisations, he explains, often operate like a multi-floor building — but AI projects are typically scoped and solved floor by floor (e.g. improving e-commerce, or campaign management) rather than considering the building and organisation as a whole. This siloed approach can deliver short-term wins, but eventually hits a ceiling, because sustained value requires connecting AI initiatives across departments; for example, linking inventory optimisation with the demand-generating marketing and campaign functions that drive it. In Lumo's view, reframing the strategic question at a whole-organisation level (rather than department by department) is the real starting point for lasting AI transformation.

Why Businesses Don't Have a Decade to Get This Right

Unlike earlier digital transformation efforts — which unfolded over roughly a decade, accelerated by the pandemic — Krantik argues that today's tougher macroeconomic environment means businesses can't afford the same luxury of time with AI. Without the growth conditions of the 2010s to fall back on, the speed at which organisations can adopt AI to drive results will be decisive, whether growth or cost savings. This drives Lumo's "two-speed" approach: pursuing a long-term strategic vision while deliberately delivering incremental wins that build both commercial momentum and workforce confidence along the way.

The Real Blockers to AI Transformation

Beyond technology, Joe and Krantik identify several recurring blockers to successful AI adoption:

• Cultural and behavioural resistance — a lack of organisational openness to test, learn, and sometimes fail
• Misalignment between long-term vision and incremental plans
• Uneven "collective readiness" — some organisations excel in one or two of Lumo's five pillars (like data or technology) but lag significantly in others, particularly workforce readiness
• A generic, poorly understood definition of "AI" itself, often skewed by the recent dominance of LLMs and conversational AI tools, obscuring the many other forms AI can take
• Short-term investor and City pressures, which push leadership toward smaller point solutions rather than sustained, enterprise-wide transformation

Real-World Case Studies: Healthcare and Footwear

Krantik shares two contrasting client examples. In a healthcare engagement, Lumo helped define a long-term customer engagement and support strategy, deliberately building AI maturity incrementally rather than assuming everything could be "agentic from day one." In a footwear and apparel business aiming to hit specific digital sales targets, Lumo had to deliver the harder message that the organisation was far from AI-ready, but found immediate value simply by restructuring engineering and product teams, cutting development team headcount by 30–40%, proving that AI readiness itself can generate savings even before deploying any AI technology.

A third example, with a large advertising services company, focused entirely on driving adoption of tools the organisation had already invested in — lifting usage from around 15% to 30–40% of the global workforce — demonstrating that meaningful AI value often comes from culture and process change, not new technology.

What's Next: Agentic AI, Governance, and Trust

Looking ahead, Joe highlights the rapid maturation of agentic AI as a major development to watch, though adoption will hinge on organisational readiness and governance. Krantik points to growing clarity around AI governance, trust, privacy, and data security as a critical unlock — arguing that many organisations are currently held back not by lack of ambition, but by uncertainty over what's permissible, and that clearer regulation and formalised governance will give businesses the confidence to move forward.