# 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._
**Published:** 2026-07-23
**Source:** https://www.startuphub.ai/ai-news/technology/2026/fda-s-ai-platform-hits-85-staff-use
---
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.

Fragmented AI EffortsDriver

each FDA center operated its own AI capabilities, leading to duplicated costsFrom the article 2 mentionsThis 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.addressed byELSA AI PlatformCoreFDA launched ELSA, an enterprise AI platform, to unify fragmented effortsFrom the article 4 mentionsFood 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.Halo Data FoundationCoreFrom the article 2 mentionsThis 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.Custom AI AgentsEffectplatform enables staff to create custom AI agents for specific tasksFrom the article 2 mentionsFDA medical doctors, scientists, and administrative staff are now independently building custom AI agents.enablesUnified Data AccessEffectbrought 50 to 60 data sources from all centers onto a single platformFrom the article 2 mentionsPreviously, each FDA center operated its own AI capabilities, leading to duplicated costs and a lack of unified data access.Improved Data SharingEffectFrom the articleData sharing improved dramatically, shifting from multi-day processes to near real-time streaming.85% Staff AdoptionOutcomeFrom the article 3 mentionsFood 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.

## 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](https://www.databricks.com/blog/how-fda-built-ai-platform-85-its-staff-now-use-daily) 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.
---
Original analysis from [startuphub.ai](https://www.startuphub.ai), the #1 AI startup directory.