Federal agencies are set to spend over $3.3 billion on artificial intelligence in FY 2025, according to NITRD reporting. But most contractors still struggle to position themselves for this opportunity. It's not just about having a tool that works. Agencies want vendors who can meet mission goals with secure, ready-to-deploy solutions.

Since 2023, we’ve tracked more than 200 AI-related awards. What separates winners from those who don’t get past evaluation is not just compliance. It's whether you can prove that your AI is technically sound, aligned with agency priorities, and ready to scale. GSA’s August 5, 2025 announcement added OpenAI, Google, and Anthropic to the Multiple Award Schedule. This expands access to generative AI tools for federal buyers. But the shift favors firms that are already aligned to current procurement trends. Here’s how to make sure you’re not left behind.

Why Most AI Proposals Fall Short

Agencies are under pressure to adopt fast, not wait for perfect paperwork. And while recent OMB memos outline guidance for responsible AI, most solicitations are still being awarded to firms that show technical and operational readiness.

Common issues we’ve seen:

• Proposals include boilerplate cybersecurity plans without addressing AI-specific risks like adversarial attacks or model drift

• Governance language is too generic and doesn’t assign responsibility or outline retraining procedures

• Ethical AI policies are copy-pasted from corporate documents and don’t explain how risk is managed in a government setting

OMB memo M-25-22 applies to high-impact AI systems beginning with solicitations issued after October 1, 2025. But even now, agencies are expecting more than marketing claims. They want trusted results.

What Agencies Actually Care About

Each agency is interpreting policy differently depending on its mission. Here’s what we’re seeing in practice:

• The Department of Defense has already awarded $200 million in AI contracts to major players with proven capabilities and security infrastructure

• DHS has emphasized human-in-the-loop design for systems that involve public safety or surveillance, consistent with earlier guidance from M-25-21

• GSA leadership has stated their intent is to reduce procurement delays, not increase regulatory complexity

This means your strategy needs to be targeted. Contractors who rely on compliance alone are getting outpaced by those who offer deployable AI with real use cases.

What the GSA Generative AI Update Means for You

On August 5, 2025, GSA added OpenAI, Google, and Anthropic to the Multiple Award Schedule. This gives civilian agencies streamlined access to generative AI tools. But that does not guarantee more awards. It means the competition just got more credible.

What changes:

• Buyers can now purchase widely used generative AI platforms under pre-negotiated contracts

• Schedule holders can team with small businesses offering custom AI services or integration layers

• Contractors bidding on high-impact AI use cases after October 1 will be evaluated under M-25-22 principles

What doesn’t change:

• You still need to prove technical alignment with the agency’s mission

• Your proposal must reflect real risk awareness, not just compliance checklists

• If you’re offering AI for the first time, you must back it up with training, oversight, and support

This is not just an opportunity for large primes. Subcontractors with niche capabilities, sector expertise, or clean compliance records have a chance to play a key role in upcoming awards.

What Actually Wins AI Contracts in 2025

We’ve worked with firms who entered the AI space through GSA teaming, agency pilot programs, and simplified acquisitions. What they had in common wasn’t just clean registration. It was clarity.

The winning proposals focused on:

• Real-world applications that tie directly to an agency problem

• Documented AI governance that fits the scale of the project

• Clear assignment of roles for auditing, retraining, and model control

• Technical maturity that meets security requirements and avoids over-promising

One client positioned their AI dashboard as a decision support tool for federal HR offices. They didn’t lead with buzzwords. They showed how it fit into the agency’s workflow. That approach helped them secure a BPA teaming role within 45 days.

What’s Next

You can try to reverse-engineer memos like M-25-22 and M-24-18 on your own. Or you can partner with a team that’s already built winning AI strategies for federal contracts.

We help contractors translate AI capabilities into agency-ready offers. From teaming opportunities to responsible use statements, we know what the buyers are actually looking for.

Talk to an expert about preparing your AI solution for government contracts

FAQ

Do I need a GSA Schedule to offer AI to the government?
No. But teaming with a Schedule holder or securing your own contract vehicle will make your offering easier to purchase, especially now that generative AI vendors are available.

What is the NIST AI Risk Management Framework?
It’s a voluntary guidance document used by many agencies to assess the safety, governance, and reliability of AI systems. It is not mandatory, but highly recommended.

Can I offer generative AI tools like ChatGPT to federal buyers?
Yes, if you can demonstrate responsible use, transparency, and alignment with agency missions. GSA access now makes this easier, but buyers still need assurance.

What are the compliance rules for different agencies?
DHS prioritizes oversight and human review. DoD focuses on secure deployment. HHS looks for fairness in health data applications. You must align your solution to each agency’s actual use case.
View full FAQ page

Top Articles

The 17 Most Common Types of Government Contracts Explained

Writing a Winning Capabilities Statement in 2025

Understanding Federal Set-Asides