London Tech Week: Astrazeneca’s AI push could take humans out of the loop
Astrazeneca unveiled an ambitious overhaul of its manufacturing and supply operations driven by deep tech, AI, and what it called a “self-healing” supply chain at London Tech Week.
The firm claimed it’s laying the groundwork to become one of the top 10 most innovative and sustainable supply chains globally by 2030.
But while the vision is ambitious, questions remain on feasibility and the human cost of automation.
“We’ve set our sights really high”, said David Baxter, head of global operations strategy and transformation. “Our ambition is to revolutionise manufacturing and supply using AI”.
Aztrazeneca, a pharmaceutical giant with $54bn (£40bn) in revenue in 2023 and operations covering 117 countries, aims to launch 20 new medicines globally in the next few years.
To support this pledge, it’s betting heavily on advanced tech like digital twins, low-latency computing, and generative AI.
“We want to ensure our innovative science reaches patients as fast as possible”, Baxter announced at the event. But, the complexity of that mission raises concerns.
Astrazeneca’s ‘self healing’ supply chain
The health titan’s concept of a “self-healing supply chain” relies on predictive analytics, large scale automation, and generative AI agents working independently – potentially even without human oversight.
“AI senses issues, demands changes, disruptions, and simulates scenarios to mitigate risk”, the firm announced. “Eventualy, we could take the human out of the loop when it comes to suggesting actions, making resolutions…”
“As we build our trust model, we can take away the human and allow our real time to mention, resolve, monitor and progress”.
That notion of AI driven decision-making without human input, triggers concerns in both ethical and operational circles.
While Astrazeneca’s team emphasised that trust and model reliability would come first, some may ask whether regulators or patients are ready for such an approach.
Digital twins and data infrastructure
The firm is already using digital simulations, otherwise known as digital twins, to model production processes, and simulate factory workflows, amongst other things.
Astrazeneca claims that this reduces the number of chemical reaction steps, lowering solvent use whilst speeding up manufacturing.
It is also applying AI to automate documentation for regulatory filings across countries using large language models.
Yet, data is a repeatedly mentioned issue, as building AI into data, the team recognised, “means a lot more AI hacking”.
That’s also a tall order in global pharma, where data silos and fragmented infrastructure remain persistent obstacles.
The AI payoff – cost cuts and faster medicine
The company’s deeper tech agenda hinges on building AI agents that can mimic human thought and collaborate – an advanced concept that researchers across the industry are still grappling with.
“We’re truing to learn about agentic architectures – how to mimic human decision-making with AI agents”, said Baxter. “This is not one agent, but a system of multiple agents synchronising with one another”.
The firm is also working on edge computing solutions to reduce latency and energy use in its production facilities.
If its strategy works, the potential rewards are large. Faster drug development timelines and reduced environmental impact could reshape how modern pharma operates.
“By 2030, this will reduce our manufacturing costs significantly”, the team explained. “That also means more money for science – and faster medicine to patients.
The firm also pointed to sustainability goals, saying it’s committed to becoming carbon neutral and reducing water storage.