Building the Foundation for Agentic Safety

Nov 13, 2026 | David Kolosic

So many safety professionals that I talk to are excited about AI. But there is a key barrier standing in the way – Legacy point solutions with siloed data and disjointed processes. Without the right foundation, AI can become a compliance liability rather than a productivity accelerator.

At World Patient Safety Congress, Sandoz’s Aniket Agarwal joined me to share his vision for unified and connected safety that gives humans, advanced automation, and AI the right foundation for innovation. Sandoz has already eliminated more than 50% of its core safety systems by implementing Veeva Safety Suite.

Read on to learn how Sandoz and other safety organizations are driving safety transformations to improve efficiency, productivity, and compliance.

Unifying and connecting safety processes, content, and data

Over time, isolated point solutions for specific PV needs have created complicated and disjoined IT patchworks that block cross-functional visibility, increase maintenance burdens, and slow response times. Moving to a truly unified cloud platform provides greater opportunity to connect processes, content, and data both within PV teams and across the organization.

How Sandoz harmonized their safety landscape

Aniket Agarwal said that a spin-off from Novartis gave Sandoz a blank sheet, and an opportunity to consolidate their R&D and quality functions on one single platform. Maintaining numerous point solutions across safety, clinical, regulatory, and quality domains creates challenges, especially when the systems are evolving at different speeds. By consolidating into a single ecosystem, Sandoz enables sustainable support for its operations.

“We want to look at how we can harmonize data and technology across global development to be able to leverage each other better,” Aniket told me on Season 1 of the Safety [R]Evolution podcast. He emphasized that manually obtaining information from clinical and regulatory systems or through complex integrations creates unnecessary maintenance. With a unified platform, all solutions grow at the same pace, giving all departments the opportunity to align and collaborate.

He echoed the same sentiment during the keynote at World Patient Safety Congress, sharing that Sandoz successfully reduced its number of safety systems by more than 50%. By implementing the Veeva Safety Suite, Sandoz brought together case processing, reporting, signal detection and PSMF on one single platform, and decreased its technical overhead and system maintenance while establishing a single source of trusted information to leverage across PV and other domains.

For example, Sandoz connected its existing Veeva RIM application with Veeva Safety to create a single, common library for product and registration information. With over tens of thousands of registrations present in over 100 countries, the impact and efficiencies gained by maintaining only one product library is substantial. This also benefits downstream processes, including case processing, signal management, literature management, and reporting & analytics, highlighting a compelling use case of automating the generation of pharmacovigilance system master file (PSMF) Annexes H, and inclusion in relevant PSMFs.

The full Veeva Safety Suite was implemented within 15 months, which is an impressive feat. This was made possible by following five guiding principles, which was helping us in the decision-making process when designing their future technology landscape:

1. Prioritize pragmatism over perfection
2. Adopt industry best practices where no fit‑for‑purpose Sandoz process exists
3. Favour configuration over customization
4. Leverage automation fully, using a risk‑based approach to GxP
5. Fast‑track deployment of technology capabilities when no current solution exists

How UCB standardized and simplified across safety operations

Other biopharmas are also prioritizing a unified safety approach. During my conversation with Adrian Maynier at UCB, he noted that UCB previously had 17 different systems just for safety operations, all with different suppliers and infrastructure, creating a complex patchwork. Maintaining multiple systems complicates not just the technology landscape, but also operational processes within safety and across other functions. This can be significantly resource and cost intensive, when all of the data could be available in one place and easier to maintain.

As a result, UCB brought case processing, reporting, signal detection, and safety content into one unified solution. Adrian said, “With more simplified and standard processes, you have this specialization and concentration of knowledge, and that makes the business more efficient and effective and leads to better quality outcomes. I’m a big fan, in that sense, of the harmonization and centralization of activities.”

Improving efficiency and productivity with automation and AI

A unified and connected platform also creates the right foundation for automation and AI. Many biopharmas are taking a phased approach to technology adoption, starting by implementing rules-based automation for high-volume, repetitive tasks before attempting to layer on AI. A clean data foundation is necessary to gain the most value from advanced automation and AI tools.

Aniket noted that if you can solve a problem with simple automation, you don’t need to look at more complex automation or AI. Before considering how to apply AI agents to tackle a particular challenge or workflow, teams should investigate the gap in efficiencies and take a pragmatic approach:

• Begin with a standardized data foundation: Standardize data across safety and other business functions to optimize automation and AI outcomes and prevent downstream errors.
• Consider rules-based automation first: Leverage deterministic, rules-based logic for high-volume tasks, before layering on advanced algorithms.
• Maintain risk-based oversight: Deploy AI tools with strong governance, validated workflows, and “human-in-the-loop” oversight to maintain regulatory compliance.

The art of the possible for AI and automation use cases

Because PV has long leveraged fragmented systems and manual processes, there is ample opportunity to improve efficiency with automation and AI. Aniket shared some of the areas where Sandoz is exploring and prioritizing potential use cases, including adverse event collection and intake, case management, signal detection and management, aggregate reports, risk management, and other areas related to PV compliance.

Aniket mentioned the significant opportunity to streamline the intake of unstructured adverse event information, as approximately 60% of information coming in is not in an E2B format. AI and automation capabilities could be leveraged to ingest and structure this data, evaluate its conformance with ICSR reportability guidelines, create a case, and advance it through case processing workflows.

Veeva Falcon Safety is an agentic workforce for safety operations that can simplify these processes by using agents to do the work. Falcon Safety will initially focus on case intake, and follow-up management to increase efficiency and reduce costs for biopharma companies of all sizes.

Shifting to strategic safety

By establishing a unified foundation and automating routine tasks, PV teams can become strategic partners to the business. Pav Rishiraj discussed the idea of inverting the operational pyramid. By spending less time on manual tasks, safety teams can dedicate more time to benefit-risk analysis and advancing safety science.

Pav said, “PV is quite often seen as a cost center, but we are now talking about strategic initiatives, and PV can generate a lot of value.” He stressed that deploying solutions that expedite decision-making helps teams make benefit-risk decisions on a rolling basis, with the goal of ensuring patients continue to receive safe and effective medication.

Redesigning lean processes

I kicked off Season 2 of the podcast with Janni Hermansen, head of safety operations at Ferring. We talked about how teams must redesign lean processes to support this technological shift:

• Discard legacy habits before upgrading solutions: Ferring mapped its case processing workflow and identified 11 handovers for a single case, underscoring that companies must simplify processes before implementing new technology.
• Get data right first time to improve patient safety: Move beyond functional siloes to approach patient safety holistically, including getting HCP and patient feedback right the first time to reduce follow up and improve outputs.
• Identify opportunities for AI: Janni explores the role of conversational AI to automatically structure natural interactions into high-quality data and compliant formats.

The future of pharmacovigilance sits with those who simplify processes first, standardize data globally, and adopt a unified platform approach to improve how humans and technology work together on the path to strategic safety.