Devia exits stealth with agentic AI and a new leadership team for pharma quality
What's the deal? Devia, a Zurich-based spinout from AI firm VisiumDealroom has a profile for this one. Try Dealroom →, emerged from stealth on November 11, 2026, with an agentic AI that manages GxP deviation investigations in pharmaceutical manufacturing. The tool is now available to large biopharma companies, contract development and manufacturing organisations (CDMOs), and contract research organisations (CROs).
Who's running it? Visium set up a dedicated leadership team for the launch: Andrea Tassistro, former chief transformation officer at Owkin; Anton Van Pamel, former partner at BCG X; and Pascal Rodriguez, former head of engineering at Bestmile.
What does it do? Devia's agent works across the deviation lifecycle — structuring events at intake, gathering evidence, proposing root causes, and drafting corrective actions (CAPA). Operators, investigators, and QA staff review and approve each step, and the tool checks root causes against past cases to flag recurrences across sites.
Why now? Inadequate investigations are among the most frequent inspection findings in the industry, and the same problems often recur across sites. Regulators also now scrutinise AI in GxP processes: in October 2026, the European Commission published EN 18286, the first harmonised standard under the EU AI Act covering quality management for high-risk AI providers.
The product grew out of Visium's work with drug manufacturers, where deviation investigations were slow, manual, and inconsistent between sites. Over the past year, the team rebuilt it as a multi-site platform and piloted it in live GxP environments, including with a large CDMO. Devia integrates with validated systems customers already run, such as TrackWiseDealroom has a profile for this one. Try Dealroom →, Veeva Vault QMSDealroom has a profile for this one. Try Dealroom →, and MasterControl.
What could go wrong? Devia enters a market where Veeva Vault QMS, TrackWise, and MasterControl are already adding AI to deviation and CAPA workflows. The company says pilot users halved investigation time, but it has disclosed no sample size or independent validation. Every AI suggestion is traceable, and a qualified reviewer signs every decision, as GxP requires.
What they said: "Devia connects them, so the real cause is found the first time and a repeat is caught before an inspector does," said Anton Van Pamel, chief product officer.
The signal: Devia's pitch is broader than bolt-on AI features: an agent that runs an entire investigation while humans keep approval authority. The real test is whether it can prove reliable, audit-ready performance and fit existing systems — not just produce AI-generated drafts.
Read more: devia.qa
Image credit: Generated with Gemini
