Presented by Red Cell Partners & Carahsoft
Government agencies looking for value from artificial intelligence should begin with a specific job and a clear way to measure success. Erin Hawley, Chief Revenue Officer at Red Cell Partners, said that approach can help agencies move beyond experiments that consume resources without producing an operational benefit.
Speaking with Fed Gov Today at the Google Public Sector Summit, Hawley described an outcome-based, shared savings model intended to connect vendor compensation with demonstrated results. She discussed a recently awarded agreement with a $100 million ceiling and emphasized that the figure was not an amount being charged to any individual agency.
Under the approach she outlined, Red Cell Partners works with an agency to identify a manageable use case and agree on what success would mean. Hawley said the company would seek a share of savings only after proving the outcome. That places responsibility on the vendor to demonstrate value from the work.
Choosing the right initial task is central to the model. Hawley pointed to activities that require substantial manual effort, contracts with cost overruns, and processes that do not stay within budget. An agency can examine those areas for a specific opportunity to reduce expense or return employee time to more important mission responsibilities.
She cautioned against trying to address every organizational problem in a single initiative. A smaller, meaningful task makes it easier to establish a common understanding of the intended result. It also gives the agency and its vendor a concrete basis for evaluating whether the approach works.
Data strategy is another consideration. Hawley advocated using information where it already resides, rather than making wholesale migration a prerequisite for progress. In the approach she described, agents connect to the existing environment and perform defined work within a designed workflow.
Her analogy was onboarding a new employee. An organization would assign that person specific responsibilities, explain expectations, and monitor performance. Hawley said agencies should bring the same discipline to deploying agents. The technology needs a defined role, governance procedures, and an understanding of how its contribution will be evaluated.
Human oversight remains part of that arrangement. Hawley said people should continue monitoring agent activity as organizations use automation to reduce the burden of routine work. The goal is to make better use of skilled government employees whose time can otherwise be absorbed by repetitive manual processes.
The interview connected acquisition choices with operational design. Agreeing on an outcome is useful only if the agency also identifies the task, accounts for implementation costs, and establishes oversight. Hawley’s proposed starting point is a job whose value the agency can explain and whose improvement it can measure. That gives both government and its industry partner a practical basis for deciding whether AI is producing the promised result.