Axios publisher Nicholas Johnston sat down with Ron Ash, Accenture Federal Services CEO, in a View from the Top conversation. The two discussed why government can move faster from testing to deploying AI, how rethinking processes and workforce integration unlock impact and what it will take for the U.S. to scale innovation.

1. First things first: What's new and exciting in the world of government?

Ash: I've been focusing on helping our federal clients move past piloting and testing AI and into the deployment phase.

• We have a deployment deficit where our agencies need to move faster to get things out in production and drive real change for citizens, veterans, and the warfighters.

Agencies are used to moving at a regular pace, but technology is moving so much faster. It's time to start rethinking how we use AI to change the way the government works and be ready for the models to improve as they go.

2. Okay, but: What has the testing journey been like as you bring these AI tools into federal agencies?

Ash: As an example, right now, we're working on a cutting-edge capability for an investigations agency that asked us to digitize their entire process for them.

To do that, we've worked with the investigators to build an agentic AI framework for AI agents to do deep research and investigation across multiple data sets, allowing the agency to scan through hundreds of hours of video so much faster.

3. What this means: What separates the agencies that are quickly transitioning from AI tests to actual use cases? What makes them more innovative than some of the folks who are going slower?

Ash: I have this general rule of thumb called the 9-to-1 rule — for every $1 that you spend on AI technology, you must spend $9 on three things:

• Getting rid of processes that don't need to exist when you have AI.
• Getting your data right — not just the data you have, but all the data that's available across different systems now that AI exists.
• Thinking about how your people are going to integrate into the system and have an agentic workforce.

These steps help agencies understand what they can use AI to do so they can focus on more important missions that require their impact.

4. The impact: With AI implementation, you start from scratch and rethink everything. Is that how you've seen the journey play out with the federal government?

Ash: Where I see government get stuck in AI pilots is trying to fix the 16th element in a 20-step process. The reality is you probably don't need 20 steps anymore.

• The most important thing to remember is we're getting out of a cycle where you spend a bunch of money for three to five years building a system and then let it run for 10 years. We're moving into an era where you're going to build a system, and then you're going to replace it in 12 to 18 months because models are getting better.

5. Take note: What makes a federal agency leader who gets it, get it? How are they figuring it out and what lessons can other leaders take from them?

Ash: The number one thing is people. You can't throw technology at this problem. You must understand how you are going to get people to accept and deploy these AI capabilities.

• As I listened to Sandeep Dadlani, CEO at Optum Insight, who was on stage before me, you could tell that he thought through this and the benefit of deploying AI through the people at a company.

They're not threatened by the work. They're seeing themselves in the future, and they're able to design AI to do the work that they want it to do. No one wants a future where the human is in the loop. We want to be humans in the lead, making an impact.

6. Looking ahead: What advice would you give to folks who work at federal agencies to help them be part of the solution and speed up the process?

Ash: First off, this is a call to action for all of us. Think about our adversaries — they are singularly focused on driving models at scale across their government.

We must do the same by:

• Coming together and uniting as a public-private partnership.
• Bringing the solutions that the government can do.
• Then, once we build those solutions, scaling them across missions with similar goals.

7. Worth a mention: Tell us a cool way AI is speeding up research in the world that you've been working in.

Ash: I hope all of you have heard about the Genesis Mission. In my mind it is the single most important thing this administration has done.

• It's bringing together the best data that our country has across the National Labs to train models to change the way scientific research is being done. I think about it every day because we're building the unified data platform that the labs can use to be able to run these experiments.

This is our singular approach to fight against our adversaries. As a nation this is our chance. We have the best models and we're putting the best data on it. But we need to take advantage of it.

Learn more.