The White House AI Action Plan: A data science roadmap for federal agencies

How the AI Action Plan helps solve federal data science’s biggest challenges

Federal agencies face a familiar problem: it takes months to deploy AI models and even longer to update them. The U.S. Navy’s Project AMMO reduced its AI model deployment time from six months to six days, and retraining time from 12 months to just two weeks. This was made possible by leveraging cross-agency collaboration and shared infrastructure. This improved efficiency allows the system to adapt to evolving threats faster. Exactly what the White House AI Action Plan, released on July 23, 2025, wants to make standard across government.

The AI Action Plan is a comprehensive strategy released by the White House to set forth clear U.S. policy goals in artificial intelligence. According to the White House AI Action Plan, “[it] has three pillars: innovation, infrastructure, and international diplomacy and security. The United States needs to innovate faster and more comprehensively than […] competitors in the development and distribution of new AI technology across every field, and dismantle unnecessary regulatory barriers that hinder the private sector in doing so.”

The plan targets the root causes: agencies stuck in silos, teams duplicating each other's work, and AI projects trapped in endless approval cycles. For federal data science teams, this means new ways to share models, access tools, and scale AI work without starting from scratch every time.

How cross-agency collaboration breaks down data silos

Cross-agency collaboration used to mean sending emails and waiting for responses. Now agencies can work together in real time while maintaining security and compliance.

The Navy's Project AMMO demonstrates this approach. Working with the Defense Innovation Unit and multiple contractors, they built integrated systems to detect underwater mines. Instead of each military branch developing separate solutions, they share validated models and testing results across authorized teams.

The White House AI Action Plan calls for this kind of collaboration government-wide. When the Department of Agriculture develops crop monitoring models, the Department of Interior could adapt them for land management. When the Treasury builds fraud detection systems, other agencies could use similar approaches for their compliance challenges.

Teams can work in secure, shared environments without compromising their individual agency requirements. They build on each other's success instead of reinventing solutions that already exist.

Accelerate scientific breakthroughs with a cloud-enabled strategy

The plan recommends investing in automated, cloud-enabled labs to speed up scientific discovery. Federal R&D agencies need platforms that can handle large-scale research and connect different teams and datasets without compromising security.

Organizations like Bayer use Domino for drug discovery and materials research. They can run experiments at scale, collaborate across sites, and maintain the audit trails that regulatory agencies require. Federal research agencies face similar challenges with even higher security requirements.

NIH researchers working on pandemic response could share computational models with CDC epidemiologists in real time. Department of Energy labs could collaborate on climate modeling with NOAA scientists. Agriculture researchers could test crop optimization models using data from multiple field stations across the country.

The platform handles the technical complexity of secure data sharing and computational scaling, so researchers can focus on scientific breakthroughs rather than IT infrastructure. When discoveries happen, they can be validated and scaled across the federal research enterprise.

How to control costs with open models

Federal agencies need AI capabilities they can control, customize, and deploy without vendor lock-in. The White House AI Action Plan emphasizes leveraging open source models precisely because they give agencies sovereignty over their AI systems.

Domino lets agencies run any models they need on government-controlled infrastructure. Teams can use Llama for general language tasks, specialized defense models for classified work, or custom models trained on agency-specific data. Everything runs through the same secure platform without per-usage fees or vendor dependencies.

This matters for both cost and security. Commercial AI providers charge premium rates for government customers and maintain control over model updates and capabilities. Open models let agencies customize AI systems for their specific missions while keeping sensitive work on government infrastructure.

The platform's FinOps capabilities give agencies real-time visibility into AI spending. Teams can see costs by user, project, or department, with automated alerts before budgets get exceeded. Resources get optimized automatically, shutting down idle workspaces and routing work to the most cost-effective infrastructure.

When the Navy's Project AMMO needed to integrate four different commercial technologies, they used Domino as the foundation. Instead of paying separate licenses and infrastructure costs for each tool, they ran everything on shared, government-controlled infrastructure while maintaining full control over their AI capabilities.

Meet federal compliance standards with secure-by-design architecture

Security requirements are integral to the White House AI Action Plan, as federal agencies can't afford AI systems that create vulnerabilities or fail in critical situations. The traditional approach of adding security afterward doesn't work when AI systems need to process classified data and make mission-critical decisions.

Domino provides secure-by-design architecture that meets federal compliance standards from the ground up. The platform runs in AWS GovCloud environments with DoD Impact Level 5 certification. Every component, from data access to model deployment, follows zero-trust security principles.

Teams can test AI systems against realistic scenarios in secure environments without exposing sensitive data. The Navy tests underwater mine detection models against expanded threat environments using real sensor data. Models get validated thoroughly before deployment to unmanned vehicles, with full audit trails that security reviewers can examine.

The platform also enables secure collaboration across security boundaries. When one agency validates an AI approach, others can access the testing methodology and results without seeing the underlying classified data. This speeds approvals because reviewers have confidence in both the technology and the security implementation. Security becomes an enabler rather than a barrier because it's built into every aspect of the AI development process.

Implement proactive and advanced governance

Automated governance workflows can reduce the time spent on AI model reviews from 11 months to just a few weeks. This is achieved by automating the manual effort of tracking every model and experimenting with full audit trails. This approach helps federal agencies manage their governance and strict compliance requirements more efficiently.

Domino automates the governance work that usually takes months of manual effort. Every model and experiment gets tracked automatically with full audit trails. The platform enforces approval workflows so agencies can review AI systems before deployment while collecting the right evidence and documentation in the background.

Policy templates adapt to different risk levels and use cases. High-risk models get thorough review processes, while low-risk analytics projects move quickly through streamlined approvals. Teams spend time on their missions instead of filling out compliance paperwork.

The same customer that spent 11 months on reviews now completes them in weeks using automated governance workflows. IT maintains all the security and compliance standards agencies require while teams can actually get their work done. This transforms governance from a bottleneck into a competitive advantage. Agencies that can deploy AI safely and quickly will outperform those stuck in manual approval processes.

Achieve AI and data science success for federal agencies

The White House AI Action Plan gives federal agencies a clear direction for AI adoption. Success depends on having platforms that can execute on these policies without compromising security or mission requirements.

Domino enables exactly what the plan envisions: agencies collaborating across boundaries while maintaining security, researchers accelerating discoveries through shared infrastructure, teams controlling costs with open models, and IT maintaining governance without slowing innovation.

The U.S. Navy's transformation from six-month deployments to six-day cycles shows what's possible when the right platform supports agency missions. The plan creates the policy framework. Domino provides the technical foundation to make it happen.

For more insights on scaling AI and data science, check out this federal civilian agency guide.