.webp)
Most organisations approaching AI adoption want to know one thing above all else: which tool wins?
Which vendor should we consolidate on? Which platform should we bet the strategy on? Which product is going to look strongest in the board presentation?
For Mark Bramwell, Chief Digital and Information Officer at Oxford Saïd Business School, that is precisely the wrong question to be asking right now. Speaking with Dr. Raoul-Gabriel Urma on the Data and AI Mastery podcast, Mark makes a compelling case for why the race to pick a winner could be the biggest strategic mistake an organisation makes.
The Case Against Picking a Winner
Mark's career has taken him across high street retail, medical research and, for the past eleven years, education. It is a sector he describes as the craziest he has ever worked in, given the intersection of academia, research, professional services, students and alumni, all operating in a devolved, siloed and deeply interdisciplinary environment. That complexity is not incidental. It is the lens through which he approaches every technology decision, including AI.
When ChatGPT entered the mainstream conversation, Oxford did not convene a committee to decide which AI tool was best. It made a deliberate, conscious decision to adopt a portfolio approach instead. Several tools, architected privately and securely within Oxford's own domains, were made available to all personas across the institution, from MBA students to faculty, researchers and professional services staff.
The reasoning is straightforward. In a landscape where the dominant AI model can move from version 3.5 to 5.4 in the space of eighteen months, making a definitive call on a single vendor is not strategic confidence. It is a gamble dressed up as conviction.
Solving the Financial Equation
The obvious challenge with a portfolio approach is cost. Giving a diverse institution access to multiple AI tools simultaneously, across thousands of users, is not cheap. And Mark is candid about this.
Oxford's answer was to move away from per-user, per-month licensing models entirely. In their place came consumption-based pricing: access for all, but payment only for genuine use. If a tool is not being used, it does not cost anything. If adoption surges, that is a problem worth having, because it signals that the tools are actually delivering value.
That shift has not been without friction. In some cases, consumption of credits and tokens has outpaced forecasts considerably. But Mark frames even that as a positive signal. High usage means high adoption. And high adoption means the tools are doing something useful, which is the entire point.
AI for Every Persona
One of the most practically useful parts of Mark's perspective is how he thinks about AI value at the individual level, rather than the institutional one.
For students, the priority is dual: equipping them with foundational AI skills for the future of work while also improving the learning experience itself. Oxford has launched an AI assistant that allows MBA students to interact with course content more efficiently, surface supplementary material aligned to their interests, and feed back data on which content is resonating and which is not.
For faculty, the focus is on content generation, summarisation and teaching preparation. For researchers, it is the ability to analyse large data sets and surface patterns in hours rather than months. For professional services staff, it is about reclaiming time, restoring work-life balance and redirecting effort toward higher-value work.
The common thread is that AI is not being deployed as a single institution-wide initiative. It is being applied with genuine nuance, persona by persona, use case by use case.
The Future Oxford Is Building Toward
Perhaps the most striking example of what Oxford's AI ambition looks like in practice is its faculty-led digital avatar. One of its professors now has an AI version of himself capable of delivering his course in any language, at any time, to anyone in the world.
For Mark, this is not a gimmick. It is a direct expression of Oxford Said's core mission: fair, inclusive, equitable access to education for every learner on the planet. AI, in this framing, is not a threat to higher education. It is the mechanism through which its most ambitious promises finally become achievable.
That is a long way from asking which tool wins.
At Cambridge Spark, we work with organisations navigating exactly this challenge: building the data and AI capability, adoption strategy and leadership confidence to turn experimentation into lasting transformation.
👉 Explore how we support data and AI transformation
