Sanofi transforms quality from reactive to predictive with in-house Plai AI

Title

Sanofi transforms quality from reactive to predictive with in-house Plai AI

Content

Sanofi's Quality Regulatory Science team is scaling Plai — an in-house AI quality-intelligence tool — from a deviation-trending pilot to a global platform that completes investigations significantly faster than the legacy hours-to-days cycle, shifting quality from reactive post-deviation response into predictive decision-making. The case draws from a presentation by Matthews, Sanofi's lead at the crossroads of regulatory science, quality strategy, and digital innovation, at the 2026 ISPE Facilities of the Future Conference. Her team acts as an early warning system for the organization, working proactively with health authorities. Asked about the biggest AI challenge, Matthews is direct: the biggest hurdle is trust. Context: Patient safety is non-negotiable, and introducing AI into quality raises essential questions — how do you validate AI models, explain outputs to regulators, and ensure algorithms don't drift? Quality organizations have long relied on deterministic, rule-based decisions; AI's probabilistic insights feel uncomfortable without context and controls. The regulatory framework is still evolving — making governance critical. Evolution: Sanofi developed Plai as a tightly-scoped pilot on automating deviation and complaint trending while generating root-cause hypotheses. Within months, the team saw faster trend identification and more targeted investigations. Matthews's formula: start focused, prove value, govern responsibly, then scale — with change management, end-user involvement, transparency, and usability embedded from the start. Architecture: Plai runs inside Sanofi's RAISE framework (Responsible AI at Sanofi Enterprise), providing guardrails for AI development, validation, and deployment — risk, ethics, and compliance addressed systematically. Two metrics underpin prioritization: the Quality Maturity Index (QMI) for internal quality KPIs, and Quality Risk Exposure (QRE) for external signals like inspection trends. AI makes both more dynamic and predictive. Operational scale: Plai is deployed across multiple Sanofi manufacturing sites globally, used by quality professionals to automate manual analysis and report drafting. By lifting routine work, investigators focus on critical thinking and risk assessment — where human judgment matters most. As Matthews emphasizes: AI is not replacing people, it is augmenting them. Future direction: Matthews predicts that within five years, AI-assisted investigations, predictive quality intelligence, and automated reporting will be standard practice rather than differentiators. She anticipates deeper integration across quality, manufacturing, supply chain, and regulatory systems — holistic digital ecosystems with patient impact at the center — alongside more harmonized global regulatory expectations for AI in GxP.