How ICF used its own AI toolset before taking it to market
Gettyimages.com / Da Kuk
Agencies increasingly want to see proof points of how these technologies work before using them, which was on the company's mind as it built out the suite.
“Show us, don’t tell us” is the government market’s standard of operations when it comes to presenting how artificial intelligence technologies can be of benefit to agencies.
In the case of ICF, that meant taking between 18 months and two years to build and use its Fathom suite of AI solutions and services for internal functions before taking them to the federal customer collective.
But ICF also wanted to showcase how the tools worked for them as proof points, Chief Technology Officer Kyle Tuberson told us.
“This started with our own need to be able to scale use cases in our organization, and so as part of that we built out the Fathom platform, which enables us to take these AI use cases to market much faster than we could before,” Tuberson said. “One that we looked at right away was contracts and administration within our organization.”
As Tuberson explained, ICF used Fathom to carry out compliance checks and evaluate the risk profiles of contracts as the company prepares responses to bids.
Internal use case number two for ICF and Fathom centered on the company’s IT organization. Tuberson said this team has built several agents to automate requests for help desk and service tickets, as well as requests for access to systems.
A third internal use case focused on market research, a core piece of the organic growth engine at all companies as that is where all contract pursuits starts from. But that is not where the application of AI stops for ICF.
“For the key aspects of proposals and all along the business development cycle, organizing resumes and project descriptions, I think we're finding we can leverage Fathom quite effectively for that too,” Tuberson said.
How has that focus on itself at first turned out for ICF on the customer adoption front? In one example, the company used Fathom to help a federal health agency develop a generative AI-powered learning system that ingests reports in different filing formats and generates summaries that feed into a dashboard.
A second example involves a different agency and its need to navigate large volumes of medical and research content. Fathom aided ICF in its design and development of a chatbot, called a “digital librarian,” to help users search publications and documents for information.
Yes, it took ICF some time fo build out Fathom and get comfortable with how it works. Also yes, ICF aims to be able to show a current or prospective customer something it rapidly prototyped within 48 hours.
But ICF has also baked that quick turnaround into its system. CEO John Wasson estimated that 80% of the company’s IT modernization work takes place in an agile framework and the “significant majority” as being fixed-price or outcome-based.
As the GovCon ecosystem is finding out, that contract structure looks poised to be more in vogue with how the Trump administration wants to do IT acquisition and larger-scale tech implementations.
“This administration is looking for you to demonstrate outcomes of value as part of your delivery quickly and if you are willing to take more of a risk in doing so,” Wasson said. “I think we’re very comfortable with that.”
NEXT STORY: Fast 50 deadline extended!