The FDA is already reviewing with AI. Most device makers' evidence wasn't built for that. Enlil's four-layer architecture grounds AI in a unified system of record, so every AI-assisted answer and agent action traces back to controlled records, permissions, and human approval.

CAMPBELL, Calif., Aug. 18, 2026 /PRNewswire/ -- Enlil, Inc., the AI-powered development traceability platform built for medical device and life sciences organizations, today introduced the four-layer governed AI architecture at the core of its platform, built to help medical device companies produce evidence that holds up when AI is applied on both sides of a submission.

The FDA launched Elsa, a generative AI tool supporting agency staff across review and compliance work. That marks a change in how submissions are examined; one that manufacturers did not choose and do not control. As reviews get faster and search across submissions gets broader, the burden moves to the company filing them: evidence assembled manually at the end of a program now must withstand scrutiny that is automated, cross-referenced, and quick.

Product data for many device companies sits in disconnected systems — quality in one, requirements in another, manufacturing and post-market elsewhere — and the connections between them live in spreadsheets and institutional memory. AI applied to that foundation produces fluent answers without traceable lineage, which is precisely what an AI-assisted review will expose.

"Every platform in this market will say it has AI. The more important question is whether that AI can be precise, governed, and provable," said Nader Fathi, CEO of Enlil. "The FDA's deployment of Elsa makes clear that AI-assisted regulatory review is no longer a future scenario. Medical device companies need a governed AI strategy on their side of the submission, one that helps them move faster while preserving the controls that make their work defensible. By unifying product data and governing AI at the architectural level, Enlil helps teams create traceable, evidence-ready records as development happens, rather than reconstruct them when an audit, submission, or AI-assisted review is already underway."

Why Bolt-On AI Falls Short in Regulated Product Development

Agentic AI adoption is outpacing confidence in it. Nearly 80% of companies have adopted agentic AI or plan to within the next year, yet 90% of data leaders report concern that AI initiatives are advancing without resolving underlying data reliability problems.¹

Many eQMS and product lifecycle management platforms have added AI as a feature layer on systems architected before generative and agentic AI existed: an assistant in one module, a summarizer in another, a call to an external AI service behind the scenes. That approach introduces three structural risks in a regulated environment: company data moving outside the environment where it is governed; polished outputs assembled from disconnected documents rather than controlled records; and agents acting without an enforced boundary between what they may read, change, or approve.

Across regulated sectors, the response taking shape is compliance by design: building validation, documentation, and oversight into how a system operates rather than attaching them as a review step at the end. Enlil applies that principle to AI governance and traceability across the product lifecycle.

How It Works: A Four-Layer AI Foundation

A unified system of record sits beneath four AI layers. Governance and traceability run through all four as enforced properties, while each layer contributes a distinct operating role:

• Enterprise Agentic Runtime. A secure, in-cloud, model-agnostic execution environment powered by AWS Bedrock. It routes each task to a curated commercial or open model based on accuracy, speed, and cost without moving the workflow outside Enlil's governed boundary.
• AI Core Services. Shared services for content augmentation, domain context optimization, agent orchestration, and workflow automation ground AI interactions in each customer's regulated content and vocabulary.
• Cross-cutting Intelligence. Read-first discovery, analysis, and synthesis span connected quality, PLM, requirements, risk, procurement, and manufacturing records. This layer can identify implications and surface inconsistencies, but it cannot alter operational data.
• Point Agents. Specialized agents perform defined tasks within tightly bounded authority. Their allowed records and actions are explicit, and compliance-critical decisions are routed for human review or approval rather than executed autonomously.

In Enlil, the connected record model is part of the platform rather than a prerequisite data project. Quality, PLM, requirements, risk, procurement, manufacturing, and post-market information share explicit relationships from the start. This breadth gives AI the context to trace the effect of a product change across functions while reducing the unsupported inference required to assemble an answer from disconnected files.

Enlil extends the same governance to connected enterprise systems. Standard REST APIs support integrations with CAD, ERP, procurement, testing, and ticketing applications, while Model Context Protocol (MCP) support allows approved AI clients to reach Enlil capabilities under Enlil's authorization and data controls.

"AI precision in regulated product development begins with a data-first architecture," said Jeremy Chone, Chief Technology Officer at Enlil. "That foundation allows us to apply one of our central principles: 'Ask Anything, Change Nothing.' AI can reason across authorized, connected records without altering operational data. When specialized agents are permitted to act, the architecture strictly limits their data access, their operations, and their approval requirements, which is how you prevent agents from going rogue while keeping every action traceable."

What It Means for Medical Device Companies

Enlil's architecture is designed to deliver five outcomes for medical device companies:

• AI operations contained within a governed environment. AI runs inside Enlil's governed cloud boundary rather than moving company data among disconnected external AI services.
• AI-assisted outputs grounded in controlled records. Because quality, PLM, requirements, risk, procurement, and manufacturing records are natively connected, outputs can be traced back to the records and relationships that support them.
• Architecture-enforced governance of access, actions, and human oversight. Read-first cross-cutting intelligence informs without altering operational data; point agents act only within defined permissions, workflows, and approval requirements.
• Faster AI-assisted work without weakening compliance controls. Teams can identify cross-functional impacts and initiate controlled follow-up work without stepping outside governed processes.
• Traceable, evidence-ready AI-assisted work. Source lineage, scoped permissions, agent activity, and human decisions can be evidenced for audits and regulatory submissions.

Educational Webinar: Precision by Design

Enlil will demonstrate the architecture in a live webinar, "Precision by Design: The Governed AI Foundation for a New Era of MedTech Regulatory Review," on Thursday, Aug. 27, 2026, at 10:00 a.m. PDT. Enlil CEO Nader Fathi, CTO Jeremy Chone, and Senior Director of Product Chiratana Pot will trace a single product change across the connected records and workflows it affects. First, read-first intelligence will identify cross-functional implications without altering operational data. Next, a bounded point agent will convert those findings into controlled action. Finally, Milestone View, Enlil's dashboard for assessing evidence completeness, inconsistencies, ownership, and approval status, will show whether the affected evidence is ready for AI-assisted review.

Attendees will leave the session with a method to pressure-test their own evidence before a reviewer's AI does.

To learn more and register, visit: Enlil Webinar Registration.

About Enlil

Enlil is a cloud-native AI-driven development traceability platform built for medical device and life sciences organizations. Designed to support regulatory readiness across the product lifecycle, Enlil connects quality, regulatory, R&D, manufacturing, and operations teams around a unified system of record. By structuring product data for traceability, auditability, and real-time visibility, Enlil helps MedTech innovators manage complexity, reduce risk, and scale compliance from concept through commercialization. Learn more at enlil.com.

All third-party product and company names are trademarks or registered trademarks of their respective owners.

Media Contacts
Christine Pearsall
Director of Marketing, Enlil
christine@enlil.com

View original content to download multimedia:https://www.prnewswire.com/news-releases/enlil-unveils-governed-ai-foundation-to-ready-medical-device-makers-for-ai-assisted-regulatory-review-302853600.html

SOURCE Enlil, Inc.