As artificial intelligence (AI) increasingly becomes part of the architecture of business, the conversation is undergoing a quiet but profound shift. Where once the headlines heralded exponential growth and ever-larger models, now, we are asking not just “how much” but “how well”, and crucially, “how responsibly”. A recent episode of the Cambridge Executive Business Insights: Rethinking AI podcast brings these questions to the fore through the lens of sustainability leadership at Specsavers, as Director of Sustainability Munish Datta shares the company’s evolving journey towards frugal, purposeful AI.
AI at the crossroads: from bigger to smarter
“What if the future of AI wasn’t bigger, but smarter?” muses Professor Jaideep Prabhu at the start of the episode, setting the tone for an exploration of AI as a lever for meaningful, accessible, and responsible change. As Datta recounts his own trajectory, from two decades at Marks & Spencer, via the UK Green Building Council, to now bridging corporate ambitions with daily sustainability realities at Specsavers, it quickly becomes clear that AI’s place in this journey is not a matter of luxury or curiosity, but urgent necessity.
Meeting the data deluge: why visibility trumps efficiency
Among the most persistent challenges for businesses with global supply chains is data: its sheer volume, complexity, and fragmentation across continents and vendors. For sustainability teams, the stakes are existential. Precise emissions data, across Scope 1 (direct), Scope 2 (energy consumption), and the notoriously elusive Scope 3 (indirect, upstream and downstream) emissions, is both a regulatory and ethical imperative. As Datta puts it, “the quality of the output is only as good as the quality of the input”.
Historically, such data collection leaned heavily on industry averages or outdated manual reporting. The magnitude and granularity now demanded, however, have outpaced human capability alone. Enter AI. Specsavers employs tools such as Watershed to synthesize massive, messy, and disparate data sets, from invoices and supplier emails to shipment PDFs, into transparent, auditable emissions reports. Rather than a black box, this approach makes supply chain emissions not just visible, but actionable: “Once we’ve built that accurate product level and supplier level data foundation, the real promise then…is that AI will be able to help us accurately forecast what our emissions will look like across our full value chain over the next five to ten years”.
Beyond carbon: traceability and the transition to circularity
But emissions accounting is only the beginning. Traceability, that is, knowing not merely the origin but the full life cycle of every component, is fast becoming the gold standard for a circular economy. To illustrate, Datta spotlights progress in the fashion industry, traditionally seen as a byword for opacity and waste. Platforms such as Tracex, leveraging both blockchain and AI, now offer “factory to shop” digital custody chains, enabling companies to map every step and stakeholder with an unprecedented degree of confidence. Imagine the power, he suggests, if such initiatives were adapted for the complex material flows of optical and hearing devices, with hundreds of global suppliers and highly variable product lifespan.
The result would be a fundamental shift: “We can stop guessing, we can move away from this sort of scattergun approach of many sustainability initiatives and start making really efficient, surgical, data driven decisions that actually move the needle on carbon reduction”. Specsavers is, in Datta’s words, “at the foothills” of this journey, beginning with building robust, accurate data streams before layering in circularity initiatives.
Business resilience in an era of climate risk
For many organisations, the urgency of climate change is no longer an abstract future but a clear and present operational risk. Flooding, extreme weather, and supply chain shocks can threaten healthcare continuity for vulnerable customers dependent on Specsavers’ services. Here too, AI has a quietly transformative part to play. By integrating climate adaptation models that predict exposures to river flooding, wind, heat and water stress down to individual facilities and project out to 2100, AI can enable real-time business continuity planning.
Instead of reacting to disaster, managers can execute pre-agreed mitigation plans mere hours before a major storm hits – moving critical stock, reinforcing logistics, rerouting deliveries, and ultimately keeping essential health services running. “This shifts us from being reactive…to proactive, which is preventing damage in the first place”. Such integration, Datta contends, is vital for enabling adaptive, resilient supply chains in a volatile climate era.
Tackling supply chain complexity: technology and human connection
One cannot discuss decarbonisation and circularity in supply chains without acknowledging their daunting complexity. Specsavers’ supplier community spans geographies, languages, digital capabilities and cultures. While AI can operate as an “ultra efficient data assistant” to categorise, clean, and flag erroneous information, human oversight remains critical: “The AI does the heavy lifting, but it’s humans, us, that validate it so that we can prove to anyone that the data has integrity”.
Furthermore, the spectre of data bias, “black box” opacity, and systemic errors is never far away. Governance frameworks and new skills, not just for tech teams, but across the sustainability and procurement workforce, are non-negotiable as AI systems embed themselves in decision-making.
The answer, Datta asserts, is not a one-size-fits-all rollout, but patient, iterative supplier engagement. Raising awareness, listening to concerns, and co-developing tailored solutions are essential to ensure every partner whether large or small, digitally advanced or more traditionally run, can come on the journey.
Skills for the new era: humans at the centre of frugal AI
A theme running throughout the episode is the recalibration of roles and skills for a future where AI augments, but does not replace, human expertise. Notable, too, is the scale of transformation: a recent LinkedIn report, Datta notes, identified “responsible AI” as the fastest-growing skill set required of sustainability professionals, with demand up by more than 500% year-on-year.
Inside Specsavers, the imperative is less about headcount reduction and more about capacity expansion: “It’s kind of an all hands to the pump situation…AI is allowing us to survive this workload challenge that we have and do significantly more and very quickly with the same amount of resource, not necessarily less resource”. The new generation of sustainability professionals is evolving from manual data entry clerks to “strategic editors”, prompt engineers, and ethical gatekeepers.
Training is hands-on and highly contextualised. For vendors and internal teams alike, the focus is on using specific, carefully chosen tools (such as Watershed and Neutrino for emissions and manufacturing intelligence) to best effect. Digital literacy is important, but so too is the ability to use, challenge, and improve the systems themselves, and to ensure supplier adoption is an ongoing dialogue.
Navigating the paradox: AI’s own environmental footprint
A less visible but no less important challenge lies at the heart of AI’s promise: its own carbon and resource cost. It’s a striking paradox: “We could be using AI to make our supply chains more sustainable, but in the process we end up increasing emissions or use of energy and water”.
Much of AI’s life cycle impact, as much as 90-95% in some cases, is generated not by initial training, but by the “inference” phase: daily use and querying of models, powered by data centre GPUs. The solution, as Datta and partners at Watershed highlight, is not abstinence but intentional frugality. Principles include:
- Right-sizing models: avoiding large, energy-intensive models for trivial tasks.
- Context minimisation: keeping prompts specific and streamlined, reducing unnecessary data transmission and compute power requirements.
- Location-conscious procurement: choosing data centres and vendors that leverage renewable energy and minimise water use.
Success here is partly a matter of metrics: tracking energy and water consumption (kilowatt-hours per square metre, litres of water used), setting aggressive reduction targets, and favouring on-site renewables wherever possible. The holy grail is regeneration: producing more energy, water or material value than is consumed, and reinvesting it back into wider systems.
From circular supply chains to a regenerative future
So what does a regenerative, intelligent, circular Specsavers look like? The vision is as ambitious as it is practical. In the near future, accurate demand forecasting (optimised by AI) will reduce stockpiles, freeing up working capital for reinvestment into more sustainable solutions. Over time, every material, from rare earths for hearing aids to spectacle hinges, can be tracked, valued not just financially but for their full environmental and social impact. This insight, Datta argues, is the missing piece for fully circular, modular design, new sharing and reuse models, and ultimately less waste.
A regenerative enterprise, then, is one in which “you’re producing more energy than you require and you’re putting it back into the grid or producing more water than you require and putting it back into the sort of water cycle”. Crucially, this vision is not in tension with commercial success – indeed, it is inseparable from it as customers, investors, and regulators alike demand demonstrable, verified progress.
Building supply chains that do more with less
At the heart of Datta’s ultimate recommendation for leaders sits the defining ethos of frugal AI: “It’s using lean, purposeful intelligence to fundamentally change and transform the way we physically manufacture, trace and move products in society…If we can master that and we can build a supply chain that’s not just efficient, but genuinely resilient and sustainable and keeps materials at their highest value for as long as possible, then we can create a more sustainable planet… and I think there is huge business opportunity from a sort of commercial, financial point of view as well”.
A more purposeful intelligence for an urgent future
AI is not a panacea, nor should it be excused from scrutiny for the problems it brings. Yet the journey mapped out by Datta and the Specsavers team offers a quietly revolutionary reinterpretation: one where digital intelligence is harnessed not for its own sake, but in service of real-world resilience, equity, and regeneration.
As sustainability professionals become AI-literate, supply chains become traceable and adaptive, and every partner in the system is engaged as a co-innovator rather than a passive recipient, the possibility emerges for business, and indeed, technology itself, to do more with less. The challenge, as ever, is to turn potential into practice: to remain vigilant to the new problems emerging even as old ones are solved, and to insist that every byte of progress is matched by moral as well as mechanical intelligence.
To explore these ideas further and hear the full conversation between Professor Jaideep Prabhu and Munish Datta, listen to the full episode of Cambridge Executive Business Insights: Rethinking AI podcast, available now, wherever you get your podcasts.
