Federal AI Needs a Workforce Plan
Not Just a Procurement Plan
Government agencies are under real pressure to adopt AI. Budgets are approved, tools are bought and pilots are launched. But the people who can run those tools are leaving and many agencies have no plan to replace them. Buying AI is the easy part. Staffing it, governing it and keeping it working is where federal programs succeed or fail.
The contract is signed. The platform is live. The announcement is made.
And then the agency discovers that nobody left in the building knows how to run it, check it or explain its decisions.
Most federal AI conversations start with the same question: which tool should we buy? It is a reasonable question. But it is the wrong first question. The harder and more important question is: who will operate, supervise and be accountable for this system once it is in place? An AI tool without the people to run it is not a capability. It is an expensive decision waiting to fail.
When the People Leave, the AI Stops Working
The Government Accountability Office (GAO) recently examined the Internal Revenue Service’s push to deploy AI. Its findings are a clear warning for every agency and contractor working in this space.
According to the IRS’s chief information officer, the agency shed about 40% of its IT staff in a single year and lost about 80% of its executives. GAO reported that the IRS wants to deploy AI, but that the workforce reductions have left capacity gaps that could undermine the effort. GAO also found that the agency does not have a plan to close its AI skills gaps. The IRS agreed with GAO’s recommendation to develop one.
of IRS IT staff shed in a single year, per its CIO
technology, data and telecom employees pushed out of federal agencies in one year
of the IRS AI governance team had left, per GAO’s review
The clearest example is in the details. An IRS official told GAO that the agency may no longer be able to use an AI model that prioritizes tax returns for audit, because it may not have enough staff to conduct those audits. Fewer audits also means less data to train or retune the model. The AI does not fail on day one. It degrades slowly, as the people around it disappear.
The lesson for every agency: AI is not a product you install and walk away from. It is a system that needs people to feed it, check it and act on what it tells them. Remove those people and the system does not stay the same. It quietly stops doing its job.
The Federal Government Is Asking Agencies to Use AI in Hiring
This creates a real tension. The government is cutting technology staff and simultaneously asking agencies to bring AI into their hiring process.
In August 2026, the Office of Personnel Management (OPM) issued a memo directing agencies to adopt AI tools more widely in recruitment. OPM said that without more AI use, agencies may undermine their hiring effectiveness and efficiency. The memo builds on the 2025 merit hiring plan, which set a goal of reducing average governmentwide time-to-hire to 80 days.
The guidance is careful on one point: AI can be used in the hiring process, but human review must stay central. AI cannot be the principal basis for a final hiring decision. Using AI for final decisions on hiring, promotions, performance, discipline or reassignments would in most cases count as high-impact use, which carries stricter requirements.
That is a useful model for any leader. The question is not whether to use AI. It is where AI helps, where humans must decide and who checks that the tools are still working correctly.
Plan the People Before the Platform
Map the Skills Each AI Use Case Requires
For every planned system, name who will operate it, who will audit its outputs and who will be accountable for its decisions. If you cannot name them, the project is not ready.
Protect the Oversight Roles First
Governance, audit and model-review staff are often the first cut and the hardest to replace. Keep them. An AI program without oversight is a compliance exposure waiting to surface.
Build Skills Before the Contract Ends
Training and knowledge transfer should be written into the program plan and funded from day one, not added after go-live when the vendor leaves and the internal team is still learning.
Measure Whether the System Still Works
Track output quality, audit coverage and decision accuracy over time. Performance that declines slowly is the hardest kind to notice, so measure it on a schedule.
- 1. Who will run this system day to day? Name the roles and confirm they are funded and filled.
- 2. Who checks its outputs? Identify the audit and review capacity needed to keep the system accurate.
- 3. Who is accountable for its decisions? A named human, not a vendor or a committee.
- 4. What happens if the vendor or key staff leave? Document knowledge and build internal capability before that moment arrives.
- 5. Does the hiring process itself need AI support? If so, define the human review points before deployment.
Procurement Gets the Tool. Workforce Strategy Gets the Result.
Federal agencies do not lack ambition on AI. They lack the people and plans to make it work after the purchase order is signed.
The organizations that succeed will be the ones that treat workforce capacity as part of the AI system itself. The tool, the people and the oversight are one capability, not three separate projects.
“A federal AI program is only as strong as the people who can run it, check it and answer for it. Buy the tool. But fund the team that makes it work.”– Tim Booker, CEO, MindFinders
Building an AI-Ready Federal Workforce?
Agencies and contractors are being asked to deliver AI capability with fewer experienced people than ever. MindFinders helps government and GovCon leaders plan the staffing, skills and oversight that AI programs depend on, before the contract is signed. Start with the workforce question. The tool decision follows.
Plan Your Workforce
