Devia Exits Stealth With AI Agents for Pharma Quality Probes

Pilot users cut deviation investigation time in half, according to Devia's own launch figures — and that number is doing a lot of work for a company that just left stealth. The Zurich-based Visium spin-out is deploying AI agents to handle pharmaceutical quality investigations while leaving quality professionals accountable for every decision.

Devia says it spent a year piloting the system with quality teams in live GxP environments — the umbrella term for regulated good-practice requirements covering manufacturing, laboratories, and clinical research. The pilots included work with a major contract development and manufacturing organization. The launch targets biopharma manufacturers, CDMOs, and CROs.

The caveat matters: Devia has not disclosed the pilot sample size or evaluation methodology. The halved-time figure is a company-reported result, not an independently established benchmark.

What does the agent actually do?

In pharmaceutical operations, a deviation is a departure from an expected procedure, process, or requirement. Investigating one can force staff to reconstruct an event from production records, standard operating procedures, attachments, and earlier cases scattered across systems.

Devia's agent structures the initial report, gathers relevant evidence, suggests possible root causes, and drafts corrective and preventive actions (CAPA). Operators, investigators, and QA staff review and approve the work as it progresses. The platform's documentation adds specifics: intake questions follow the organization's own procedures, and the system can read attached photographs and PDFs. For root-cause analysis, it drafts Ishikawa diagrams and 5-Why investigations built from the event report, procedures, and relevant historical deviations.

"Quality teams don't need another system of record. They need an agent that does the groundwork, so their experts can focus on the root causes that matter," said Pascal Rodriguez, Devia's CTO.

Those methods provide structure, not proof. The real test is whether each proposed explanation is backed by evidence a qualified investigator can examine.

How does it handle past cases?

Devia says it compares new root-cause proposals with previous cases and keeps reviewed expert decisions available for future investigations. The company distinguishes a cause that keeps recurring without effective intervention from one that returns after a corrective action was verified as effective — a distinction that changes the investigation itself. An organization may need to address an untreated systemic issue, or explain why a supposedly effective fix failed.

Humans stay in the loop. Staff accept or reject suggested relationships, and those decisions become part of the record. That review process is central to the product: an unsupported AI suggestion should not silently become the precedent for the next case.

Can it plug into existing quality systems?

The launch names TrackWise, Veeva Vault QMS, and MasterControl as platforms customers already use. Devia's integration documentation gets more specific: historical deviations and CAPAs are imported from QMS exports, procedures come in as documents, and completed investigations produce closure reports matching the customer's QMS format. That's not full synchronization. Direct connectors for quality management, manufacturing execution, laboratory information management, and ERP systems are listed as upcoming, alongside broader trend monitoring, CAPA-effectiveness features, and inspection-support tools on the roadmap.

On deployment, Devia says the platform can run in the customer's Azure or AWS account, in a chosen region, using Azure OpenAI or Amazon Bedrock. Customer records are not used to train models for other organizations, per its documentation. The company doesn't claim a proprietary foundation model; the described value lies in applying AI to company procedures, historical cases, and controlled review workflows.

What about compliance and traceability?

The stakes are established. The FDA's review of drug inspection observations identifies failures to investigate discrepancies and product failures among recurrent manufacturing-quality problems — and faster drafting doesn't fix weak evidence or conclusions.

Devia's security documentation describes individually identified users, role-based access, single sign-on, encryption, and time-stamped records of changes and approvals. The company supplies validation documentation to support customers' computerized-system validation work — support, not a substitute for the customer's own responsibilities. The launch also references EN 18286:2026, the harmonized standard the European Commission confirms addresses quality-management requirements for high-risk AI providers under Article 17 of the AI Act. Devia says its approach follows those principles; the announcement doesn't establish independent certification.

Devia's opportunity is reducing the time investigators spend collecting and organizing information while making the reasoning easier to inspect. The more consequential test is whether quality teams can consistently turn that assistance into stronger investigations — and corrective actions that stop the same manufacturing problem from coming back.