FDA's AI Platform Hits 85% Staff Use

The FDA's enterprise AI platform, ELSA, achieved 85% staff adoption in two months by breaking down data silos and enabling custom AI agent creation.

Jul 23 at 8:15 PM6 min read

FDA building exterior with 'Food and Drug Administration' text visible — The U.S. Food and Drug Administration headquarters.

Visual TL;DR. Fragmented AI Efforts addressed by ELSA AI Platform. ELSA AI Platform powered by Halo Data Foundation. Halo Data Foundation enables Unified Data Access. Unified Data Access leads to Improved Data Sharing. Unified Data Access drives 85% Staff Adoption. ELSA AI Platform allows Custom AI Agents.

Fragmented AI Efforts: each FDA center operated its own AI capabilities, leading to duplicated costs
ELSA AI Platform: FDA launched ELSA, an enterprise AI platform, to unify fragmented efforts
Halo Data Foundation: ELSA underpinned by Halo, a governed data foundation running on Databricks
Unified Data Access: brought 50 to 60 data sources from all centers onto a single platform
Improved Data Sharing: data sharing shifted from multi-day processes to near real-time streaming
85% Staff Adoption: achieved 85% staff adoption within two months of launch across 16,000 staff
Custom AI Agents: platform enables staff to create custom AI agents for specific tasks

Visual TL;DRQuickExplainDeeper

The U.S. Food and Drug Administration (FDA) has achieved remarkable adoption of its new AI platform, ELSA, with 85% of its 16,000 staff using it daily within two months of launch. This initiative, underpinned by a governed data foundation called Halo running on Databricks, aimed to unify fragmented AI efforts across the agency's eight distinct regulatory centers.

Breaking Down Silos

Previously, each FDA center operated its own AI capabilities, leading to duplicated costs and a lack of unified data access. The consolidation effort, detailed on the Databricks blog, brought 50 to 60 data sources from all centers onto a single platform.

This move was accelerated by the success of the Center for Drug Evaluation and Research (CDER), which had already established a Databricks data platform. Data sharing improved dramatically, shifting from multi-day processes to near real-time streaming.

Concerns over sensitive regulatory data were addressed by Databricks' Unity Catalog, which provided granular access controls and ensured data containment.

From Chatbot to Agentic AI

With the governed data foundation in place, ELSA was rolled out. Users access multiple AI models through a single interface, enabling rapid task automation.

FDA medical doctors, scientists, and administrative staff are now independently building custom AI agents. These agents leverage standard operating procedures and regulatory documents to answer specific, grounded questions instantly.

Accelerating Research

The impact is significant, particularly in drug application reviews. Previously, reviewers spent days sifting through millions of pages to identify starting materials for drug manufacturing.

Using Databricks ML and NLP capabilities via MLflow, this process has been streamlined. Reviewers can now input an application number and receive the required information in approximately three minutes.

The FDA is focused on extending this model across all centers. The goal is to free up review staff from information retrieval, allowing them to concentrate on core tasks of ensuring product safety and efficacy.

© 2026 StartupHub.ai. All rights reserved. Do not enter, scrape, copy, reproduce, or republish this article in whole or in part. Use as input to AI training, fine-tuning, retrieval-augmented generation, or any machine-learning system is prohibited without written license. Substantially-similar derivative works will be pursued to the fullest extent of applicable copyright, database, and computer-misuse laws. See our terms.

#FDA #Artificial Intelligence #Generative AI #Databricks #Data Governance #AI Adoption #Public Sector Technology #Machine Learning #NLP