L7 Informatics argues that governance must exist in the operational fabric before agents can act, and that no layer added on top of fragmented systems can provide it.
AUSTIN, Texas, July 22, 2026—As agentic AI moves from pilots into regulated pharmaceutical workflows, a growing number of enterprise software vendors are positioning their solutions as the "governance layer" for AI. Vasu Rangadass, Ph.D., Founder and CEO of L7 Informatics, is challenging that concept at its foundation. In a new article, he argues that governance behaves as a property of the operational environment where work is executed, and that treating it as a layer to be added on top of existing systems misdiagnoses the problem the industry is trying to solve.
The argument arrives at a moment when the regulatory direction is increasingly clear. The FDA's draft guidance on the use of artificial intelligence to support regulatory decision-making for drugs and biologics is built around a risk-based credibility assessment framework, under which sponsors are expected to establish the credibility of an AI model within its intended context of use and support it with evidence. The agency has also disclosed that submissions containing AI components already number in the hundreds and continue to increase.
According to Rangadass, the questions an inspector will ask about an AI-assisted decision concern evidence, accountability, and control far more than model performance: where the source data came from, which version of which model produced an output, who approved the change that placed an agent into a workflow, and who owns the outcome. When those answers must be reconstructed from disconnected audit logs, exported spreadsheets, and institutional memory, a deployment is one audit away from a pause.
"A layer sits on top of something, and governance that sits on top of fragmented systems can only observe fragments," said Rangadass. "Lineage, context, traceability, and accountability either exist where the work is executed, or people end up reconstructing them under deadline pressure. This industry has decades of experience with how that second approach performs during an inspection."
The problem compounds, Rangadass continues, because pharmaceutical workflows always cross system boundaries. A technology transfer alone touches development data, manufacturing execution, quality systems, and regulatory documentation. If each platform in that chain introduces its own governance model, organizations inherit multiple model inventories, multiple audit trail formats, and multiple definitions of validation, with no single layer able to see the complete operational context.
L7 Informatics designed its architecture around the opposite sequence. The company built L7|ESP® as a unified operational backbone first: a platform where scientific data, processes, instruments, applications, and people connect through a shared ontology, with lineage and traceability captured as work is executed. L7|SYNAPSE™, the company's agentic AI layer, then brings agents into workflows where context, provenance, and execution history already exist.
"When an agent operates in an environment where lineage, version history, and human oversight are native characteristics of execution, governance becomes an outcome of the architecture rather than a separate initiative," said Rangadass. "That is the difference between being AI-ready and being AI-actionable."
The full article, "AI Governance Starts Before Agents Can Act," is available at: https://l7informatics.com/blog/ai-governance-starts-before-agents-can-act/
About L7 Informatics
L7 Informatics is on a mission to accelerate the pace of science. L7|ESP®, our Enterprise Science Platform, is the operational backbone of modern life sciences organizations, orchestrating people, instruments, data, and AI agents that drive research, development, and manufacturing. Built for regulated environments, L7|ESP unifies laboratory, manufacturing, and operational workflows into a single execution layer, giving science the infrastructure it needs to move faster and comply by design. L7|SYNAPSE™, L7|ESP's agentic AI layer, brings reasoning and autonomous execution to that environment while preserving the compliance and auditability that regulated operations require.