Beyond AI pilots: Pharma needs AI-native operating models, not more AI tools

According to Stanford University, last year, across all industries, $250 billion was invested in AI. In the pharmaceutical industry alone, the AI market is projected to grow from $4 billion this year to $25.7 billion by 2030. Despite these billions invested in artificial intelligence, many pharmaceutical companies remain stuck in pilot mode, unable to translate experimentation into measurable productivity gains.

AI has shown promise in areas such as medical writing, clinical trial analysis, and regulatory documentation, but it has yet to deliver the step-change in productivity many sector leaders expected. This disconnect between hype and reality is causing confusion and frustration for decision-makers trying to navigate new terrain.

The evolving AI ecosystem

The AI ecosystem is rapidly moving from one or two frontier models to a diverse marketplace of open models, specialist models, and operational tooling.

For AI implementation to be most effective, organisations need to move away from exploring single models and toward the question, ‘Which portfolio of models and runtimes does my product need, and how will I operate them safely and cost-effectively?’.

For pharmaceutical leaders to be able to answer this question, it requires a deeper understanding of the available tools and how they can be integrated. Lightweight fine-tuning techniques and model-compression methods now make smaller or open models powerful, fast, and affordable enough for many real workloads. At the same time, a more mature tooling architecture – one that covers prompt management, retrieval pipelines, model routing, monitoring, and safety – has emerged to support multi-model operations.

However, we can’t ignore the complexity of what happens post-production. Sovereignty – the ability for an organisation to have free choice and control over the operation of its AI systems, as well as the underlying tech foundations they depend on – is critical in any organisation, and even more so in heavily regulated sectors such as pharmaceuticals. That’s why a multi-model strategy should be implemented to ensure resilience. The emphasis of the UK’s Medicines and Healthcare products Regulatory Agency (MHRA) is on ensuring AI – or any tool within the sector – is used safely, effectively, and with appropriate oversight, rather than regulating the technology itself.

With compute options broadening across cloud, on-premises, and specialised hardware, the practical imperative for businesses is clear: choose models based on the task, not the vendor, and adopt a hybrid approach where frontier models handle the hardest reasoning problems, while smaller, customised or self-hosted models deliver cost-effective, privacy-aligned performance at scale.

The pharmaceutical industry is unique in its process, legislation, and reporting requirements. With this in mind, the focus needs to shift from simply adopting a large language model, to building a cohesive system that leverages the right components for the role. This is essential for moving beyond simple experiments to creating dependable, enterprise-grade AI applications that deliver real value to the organisations using it, and the industry as a whole.

As we’ve seen recently with the US government directive that caused Anthropic to disable its models for more than a fortnight, resilience also can’t be overlooked.

In any heavily regulated sector, proposed AI and technologies face much greater scrutiny than many other industries. The emphasis is not so much on which model is used; rather, the need for it to be secure and compliant.

Expectations vs implementation reality

Many teams expected AI to be a plug-and-play solution that would instantly boost productivity. Maybe we can blame overpromising marketing; early demos made large models look like magic: give them text, they give you answers, and suddenly whole workflows seem automatable. In reality, moving from experiments to production exposes a far more complex, engineering-heavy challenge.

Arguably, the more immediate crisis right now for many companies is in the underwhelming results experienced in rolling out AI across their workforces. The common ‘scattergun’ approach of enabling Copilot or Gemini across the business, and hoping it will yield good results, is proving fruitless. Despite the billions being invested, many business leaders then draw the conclusion that AI isn’t a good fit for their company.

However, the core issue is, in fact, context. If foundation models don’t have correctly-scoped access to the knowledge of a business, myriad privacy, safety, security, and compliance requirements restrict what they can actually deploy.

Without a clear strategy to connect proprietary business data to these powerful tools, the outputs are generic and fall short of expectations. If an organisation doesn’t have a solid plan to connect its data to AI tools, the results will likely fall short.

Another crucial factor is cost. Not simply initial costs, but modelling the running cost before roll-out is critical for any organisation to fully understand the financial implications of running AI tools at scale.

Building AI that lasts

The excitement surrounding generative AI is beginning to give way to a more pragmatic phase of adoption.

Over the next few years, the gap between organisations that simply deployed AI tools and those that fundamentally redesigned how they work will become increasingly apparent. Some businesses will question the return on expensive AI licences because they never built the foundations needed to generate meaningful value. Others will realise lasting competitive advantage comes not from using AI, but from integrating it into the fabric of the organisation.

For pharmaceutical companies, that means moving beyond pilots and isolated use cases. It means building AI-native operating models that combine the right models, access to the right data, and the right governance to solve business problems at scale. In fact, McKinsey & Company’s view is that agentic AI’s benefits will be boosted through the evolution of AI from ‘tool’ to ‘coworker’ and catalysing an end-to-end reimagining of the value chain.

The winners will be those who have thought deeper about their rollout plans to deliver purposeful AI solutions grounded toward the unique data, information, and knowledge requirements of their organisation. This means moving beyond generic applications and building custom solutions that leverage a pharmaceutical business’ unique assets – including its people.

About the author

Ciaran Cosgrave is the CEO of Nearform. He is passionate about leveraging technology to empower businesses, accelerate growth, and build enduring value.