Artificial Intelligence is increasingly shaping the daily realities of how businesses interact with customers, manage operations, and drive innovation. In the latest episode of the Cambridge Executive Business Insights: Rethinking AI podcast, Professor Jaideep Prabhu explores this rapidly evolving landscape in conversation with Laura Joseph, Chief Customer Officer at Tesco Mobile. Their discussion cuts deeply into the nuances of “frugal AI,” the subtleties of customer trust, and the essential need to keep humans firmly at the centre of AI-driven transformation.
Purpose beyond power: the philosophy of frugal AI
AI promises immense capability, but scale by itself is not the hallmark of progress. This “frugal AI” ethos is less about technological austerity and more about judicious, human-centred application. By focusing on accessibility, fairness, and context, frugal AI seeks to unlock new forms of intelligence, equity, and efficiency precisely when the world most needs them.
For Laura Joseph, this means exploiting technology not for its own sake, but to enhance experiences for both customers and colleagues. At Tesco Mobile, AI acts as a force multiplier for values that are enduring: simplicity, relevance, and, above all, effortlessness in every customer interaction.
Human-centric storytelling meets data-driven insight
The route to this approach was, for Laura Joseph, an unconventional journey. Starting in classics and academia, her interests in the power of storytelling and the structure of language eventually fused with a growing passion for computation and technology. This unusual combination of narrative and data would become the foundation of her philosophy: that technology should amplify, not override, the human need for meaning and connection.
At the heart of Tesco Mobile’s customer experience strategy is the notion that AI and data are not endpoints. Rather, they are vehicles for helping people to achieve what matters to them. For example, by deploying predictive models, the business can alert customers proactively when they’re likely to run out of data, offering tips and solutions before inconvenience arises. The focus is not on extracting ever more value from the customer, but on enhancing trust and genuinely helping, which is a subtle but significant distinction.
Personalisation, is not about manipulating choices or foregrounding the mere cleverness of technology. As Laura puts it, “personalisation shouldn’t be something customers notice; it should be about us getting it right for the customer”, where helpfulness is unobtrusively embedded in every touchpoint.
Proactive support: AI as the enabler, not the star
One of the most tangible illustrations of this philosophy is Tesco Mobile’s “out of data” model. Rather than wait for customers to receive an unexpected bill or be stranded without connectivity, the AI system analyses usage patterns with techniques like velocity and propensity modelling. It can then prod the customer at just the right moment, not only advising on top-ups but highlighting app behaviours or suggesting practical adjustments, depending on what the data reveals.
This is proactive rather than reactive care, something few companies have mastered at scale. The intelligence is in the anticipation, not in the after-the-fact clean-up.
Importantly, the communication methods themselves are guided by humility: the simplest channel (in this case, a timely text message) is used when the situation is urgent or when it matters most. In an ecosystem where digital noise is constant, relevance and restraint become defining virtues.
Differentiation beyond commoditisation: competing on experience
The mobile telecoms market is notoriously competitive and, to a large extent, commoditised. Pricing parity and product similarity make it difficult to stand out. So what room is left for innovation? For Laura, differentiation comes not from ever-escalating offers, but from becoming a partner to the customer, by preventing inconvenience, building trust, and always prioritising helpfulness over purely commercial goals.
This is a disciplined use of AI; one that prioritises outcomes over outputs, and rigour over “dashboards that change absolutely nothing for our customers”. Data, in this approach, is not an ocean to drown in, but a map guiding purposeful decisions.
In practice, this means starting all projects with the desired decision or customer outcome, not with the technology. Unless Tesco Mobile can clearly articulate what will change, who will own the change, and why it truly matters to the end user, the initiative does not proceed.
Modernising data and operating models: insights to action
Transforming insights into actions is easy in theory, but fiendishly difficult in practice, especially in large organisations with legacy systems and “parallel versions of the truth”. Tesco Mobile is investing heavily in modernising its data infrastructure, ensuring that teams have unified, trustworthy sources of truth.
Critically, the link between insight and action is institutionalised; operating models ensure every insight is owned, measured and acted upon, with AI even helping to prioritise which improvements matter most.
One example of insight-led change is the overhaul of the inbound voice response (IVR) system. When a new AI-powered natural voice system was trialled, the team noticed a small subset of customers failed multiple times in a short span, which was a sign the system wasn’t serving everyone equally. Rather than chalk these up as operational successes or blame the customers, Tesco Mobile changed tack, bypassing the IVR entirely for those who struggled repeatedly, delivering them directly to human agents. This kind of nuance, and willingness to revisit previous assumptions, epitomises human-led AI, one that listens and evolves.
Governance, trust, and the role of humans
Joseph argues for human accountability over algorithmic absolution. Even as AI supports ever more decisions, ultimate responsibility for outcomes always remains with people.
The principles guiding Tesco Mobile’s AI practice of transparency, human accountability, and compliance/security by design, are not just tick-boxes, but foundational to earning trust. This means being explicit both internally and externally about when and how AI is used, and making it easy for customers to access human support whenever needed.
A distinguishing feature of the company’s approach is the deeply embedded AI governance model. Every AI use case, tool, or experiment must pass through a multidisciplinary governance forum, composed not just of technologists but broader business experts. No experiment is rolled out at scale without rigorous scrutiny: legal, privacy, and risk concerns are weighed alongside customer outcomes.
Rather than slowing progress, this model (and an accompanying AI Guild, an inclusive, self-selecting learning community) noticeably speeds it up, because risks are surfaced and managed in advance, not locked away to derail initiatives months down the line.
Culture change, experimentation, and organisation-wide literacy
Perhaps the most profound shift underway is cultural. AI and data are recognised as organisational, not technical, transformations. All staff are encouraged and required to invest in developing AI literacy: experimentation is celebrated, failures are shared and learned from, and every team is charged with uncovering and reporting their own use cases.
Training is not a perfunctory task, but a ramp for bottom-up innovation. Legal teams and finance analysts are discovering their own ways to use Copilot and other AI tools; the resulting ideas are more impactful than any top-down directive could hope to be.
The entire approach is agile by design: teams are empowered to run controlled experiments, test assumptions rapidly, and, just as importantly, roll back features or interventions if they aren’t delivering value or raise ethical concerns. This “try fast, scale only proven successes” mantra is central to a sustainable AI strategy.
Empowering colleagues for better service
Where many organisations focus AI exclusively on customer-facing problems, Tesco Mobile places the same emphasis on internal empowerment. Store colleagues and frontline contact centre agents are all equipped with next-generation tools, information, and intelligent prompts to help them excel, especially in difficult, emotionally charged situations such as bereavement or vulnerability. AI, then, becomes a lever for upskilling and augmenting colleague capability, not displacing it.
Integration with the broader Tesco ecosystem offers additional sources of customer value, such as Club Card points, combined offers, and a unified digital life, further amplifying the advantages of a joined-up, data-driven approach.
Metrics, meaning, and measuring impact
The metrics used to evaluate AI’s success are as nuanced as the organisation’s goals. For Laura Joseph, the customer lens is threefold: performance (did things get better or easier?), behavioural impact (did avoidable contacts go down or journeys improve?), and perception (did customers truly feel helped and respected?). These measures are balanced against cost, efficiency, and internal satisfaction to paint a genuinely multidimensional picture.
Yet, the company moves cautiously: every rollout remains in test-and-learn mode, always mindful of latent bias, unexpected harm, or unanticipated shifts in customer sentiment. Truly responsible AI is measured as much by what it chooses not to do as by what it accomplishes.
The future of AI-driven relationships: augmentation, not automation
Looking ahead, neither Laura Joseph nor Jaideep Prabhu predicts a future where AI renders the human touch obsolete. If anything, customer expectations for seamless, responsive and personal service will only rise. The best AI will recede into the background, quietly removing friction and empowering people (both customers and colleagues) to thrive.
There will always be appetites for different engagement modes: some users will crave full self-service, others will seek reassurance and empathy from human agents, and many will drift somewhere in the middle. Building effective systems, therefore, demands both technical excellence and a deep respect for the irreducible complexity of human preference.
The analogy to the evolution of internet banking is inescapable: as with the shift from bricks and mortar to digital channels, AI’s impact will most likely materialise as a mosaic, with old and new co-existing, serving different needs at different times.
Practical advice: start small, govern thoroughly, and put people first
What should other businesses take from this case study? In a world where AI is “the next big thing,” Laura Joseph cautions against indiscriminate adoption. Instead, leaders are urged to:
- Design for real customer needs, not the seductions of the latest technology.
- Start with safe, small-scale trials, preferably by experimenting with internal colleagues before full public releases.
- Build robust governance and risk processes from the outset, not as afterthoughts.
- Encourage accountability and ensure people, not algorithms, remain responsible for decisions.
- Speak plainly with customers about when, where, and how AI is deployed, and always guarantee access to empathetic, human support when it’s most needed.
This is frugality, not in the sense of pinching pennies but as an ethos: stripping away the unnecessary, experimenting without hubris, and placing human dignity and truth above all.
The road ahead: from technology to trust
As AI adoption gathers pace, leaders everywhere are called upon to balance innovation with oversight, and ambition with humility. The most successful AI-powered businesses will be those that embed their technology quietly, transparently, and with steadfast care for the people they serve.
To hear the full conversation between Jaideep and Laura, listen now, wherever you get your podcasts.
