This Week on Life Sciences Digital
Big Pharma Keeps Building the AI It Used to Buy From CROs and Biotechs
Three separate deals this week point to the same shift: large pharma companies are putting AI-driven capability inside their own R&D organisations rather than only licensing it externally. Astellas' chief R&D officer described a multi-year effort to internalise clinical development, cutting out CRO middlemen and using AI-assisted tools for tasks like protocol drafting and translation; the broader cost transformation behind this has saved the company roughly $406 million over two years, against a three-year target of around $940 million. Separately, Sanofi and Owkin extended their five-year AI collaboration to build autonomous drug discovery agents on Owkin's K Pro platform, directly building on last week's news of Sanofi licensing K Pro as an AI scientist. And Novartis signed a new multi-year deal with Orionis Biosciences worth up to $1.4 billion in milestones (plus $40 million upfront), pairing Orionis' Allo-Glue molecular glue platform with its AI-driven discovery engine to find drug candidates for previously "undruggable" targets.
None of these are AI vendors selling software into pharma. They're pharma companies either building internal AI capability (Astellas), co-developing it with a long-term partner (Sanofi/Owkin), or paying for access to someone else's AI-native discovery engine on a royalty-bearing basis (Novartis/Orionis). Three different structures, same underlying logic: AI capability in drug discovery and development is becoming something pharma wants to own a piece of, not just consume as a service. For vendors selling point solutions into this market, the bar is shifting from "useful tool" to "something we'd want embedded or co-owned."
Generated with NotebookLM
Also Last Week:
Microsoft secured an agreement to roll out Microsoft 365 Copilot, Copilot Studio, and agent-governance tooling to over 505,000 NHS England clinicians and support staff. The deal follows a pilot across 90 NHS organisations involving 30,000 workers, who reported saving an average of 43 minutes per day on administrative tasks. NHS England plans to onboard 200,000 users within six months and the full workforce within a year, and Copilot will not be permitted to make clinical decisions — any AI-generated content entering a patient record must be reviewed and signed off by a registered clinician.
For a system serving 56 million people and employing 1.4 million staff, a 36% workforce rollout is infrastructure, not a pilot. It also gives the industry its first large, government-published benchmark for AI productivity gains in a clinical setting — a number vendors elsewhere will likely start citing.
Tata Consultancy Services became a Global Premier Partner in Anthropic's Claude network, giving 50,000 employees across 56 countries access to Claude and building a dedicated business unit for Claude-powered solutions in regulated industries including life sciences, healthcare, financial services, and medtech. TCS's UK life and pensions arm, Diligenta, will use Claude to support more than 22 million policyholders, and TCS engineering teams will contribute reusable skills and plugins to the Claude Code ecosystem.
Combined with the NHS deal, the week shows two of the largest AI distribution agreements in healthcare and life sciences landing within days of each other, both pitched on the same premise: governance and auditability as the entry ticket for regulated sectors.
Tool Spotlight from our Life Sciences Digital database
SOPHiA DDM™ for Multimodal
REAL-WORLD DATA & MARKET INTELLIGENCE
A platform that combines genomic, radiomic, and clinical data into multimodal AI analysis for oncology, supporting trial intelligence and real-world data workflows across pharma, biotech, and diagnostics organisations.
🚀 Your solution, in front of the people building the future of life sciences.
🎯 Reach over 30,000 professionals. Starting from €999.
Signals & Market Moves
Andhra Pradesh Runs Two Parallel AI Health Pilots at Once 🔗
Andhra Pradesh launched an AI pilot for diabetic retinopathy screening across government hospitals in Guntur, Kurnool, and Visakhapatnam, using AI to analyse fundus camera images and flag patients needing urgent referral — about 9,000 patients are expected to be screened over three months. In parallel, the state is piloting Shishu Maapan, an AI tool built by Wadhwani AI that estimates a newborn's weight, length, and head and chest circumference from a short smartphone video, removing the need for physical measuring equipment during home visits by ASHA workers.The signal: Two unrelated AI tools, same state, same month — that's not coincidence, it's a state health department actively running multiple AI pilots in parallel rather than waiting for one to finish before starting the next. Both tools follow the same pattern: take a task that previously required equipment and a trained specialist, and make it possible with a smartphone and a trained AI model. For companies building diagnostic or screening tools aimed at low-resource settings, India's state-level health departments are clearly willing to move fast on pilots, which makes them worth watching as an entry point.
The EMA published research priorities for AI across the medicine lifecycle, based on a survey of 273 stakeholders, with system robustness and explainability topping the list ahead of data governance and bias. Separately, the UK's MHRA launched a regulatory sandbox where pharmaceutical companies can test AI tools for medicine safety and efficacy directly alongside regulators, with an initial phase covering up to five methodologies and an explicit link to reducing animal testing.
The signal: Robustness and explainability outranking data governance suggests European and UK regulators are most worried about AI tools that work in testing and then degrade or become opaque once deployed on real data — not about the data itself. Combined with the Microsoft NHS deal this week, which already builds in mandatory clinician sign-off for anything AI-generated entering a patient record, the direction is consistent: large-scale AI deployment in healthcare is proceeding, but regulators on both sides of the Channel are moving in parallel to define what "proceeding responsibly" actually means before it gets harder to walk back.
Events & Calls
BIO International Convention 2026 — San Diego, June 22–25
The biggest biotech industry gathering of the year returns to San Diego next week. With AI-driven drug discovery now a standard agenda item across exhibitors and panels, expect plenty of announcements timed to the event — companies like Evogene are already confirming presentations on their AI platforms.
The Bioprocessing Summit — Boston, August 10–13
Focused on cell and gene therapy manufacturing and commercialization, an area where 167 tools in our database fall under Drug Discovery & Molecular Design — worth flagging for companies in that space.
ESC Congress 2026 — Munich, August 28–31
The world's largest cardiology meeting is built around a "Spotlight on Artificial Intelligence" theme this year, covering AI as a co-pilot across diagnosis, treatment, and clinical workflows — a strong signal of how fast AI is becoming standard in cardiovascular care.
🤝 Got a tool for life sciences you'd like more people to know about?
Submit your tool here.
Want to be featured in a future issue or explore sponsorship?
We highlight the AI tools, companies, and initiatives shaping the future of life sciences. If you'd like to collaborate, reply to this email or get in touch: [email protected]
