PEO Enterprise invigorates Army SBIR program with “pretty overwhelming” AI/ML response

The U.S. Army’s Small Business Innovation Research (SBIR) program has been around for some time, but Program Executive Office (PEO) Enterprise recently created some buzz with its initial foray into small-business solicitation. 

In 2023, PEO Enterprise’s Acquisition Innovation Directorate (AID) talked with the Army SBIR office about its idea to leverage artificial intelligence/machine learning (AI/ML) technologies to help speed up the acquisition process, which it described as “broken.” 

“Federal acquisition takes too long — particularly DOD acquisition,” said Patrick Colleran, chief of the Contracts Management Division at PEO Enterprise. “It can take seven to nine months to get a simple contract action.”  

In a near-peer threat environment in which quick action is paramount, that type of delay can be costly. 

One of the many delays in the acquisition process, said Colleran, can be the acquisition requirements package (ARP), the collection of documents that demonstrate what the government is buying and how applicable laws and regulations are satisfied.  

According to the ARP Handbook, an ARP routinely consists of a variety of documents, including a performance work statement with a performance requirement summary, a quality assurance surveillance plan, independent government cost estimate, proposal evaluation factors and a DD Form 254 – DOD Contract Security Classification Specification, though the exact collection of documents depends upon the specifics of the particular action. It is not uncommon for an ARP to include documents that are over 100 pages, said Colleran.  

As the DOD hones its capability requirements, the content of ARPs often changes as well. This can pose a challenge to contracting teams in charge of reviewing them. Often, a lot of back-and-forth ensues between contracting teams and requirements developers as ARP reviewers try to trace changes throughout the documents and ensure they don’t conflict with other parts of the acquisition package. 

“We lose a lot of time teasing out inconsistencies,” said Colleran. “But if we don’t do this, they will be on full display in the solicitation.” 

Discrepancies in acquisition packages can lead to more delays during the industry Q&A process and potentially to contract protests down the line.  All this hinders the DOD’s ability to promptly deliver critical capabilities to warfighters. 

“Since all this stems from the challenge of individuals tracking changes, not just through a single document but through the complete acquisition package, we want to use AI/ML to do that and accelerate the path to contract award,” said Colleran. 

The Army SBIR Office’s AI/ML transition broker team agreed with PEO Enterprise’s pitch and agreed to fund up to three ASA(ALT) “Direct to Phase II” awards valued at up to $2 million over a 24-month performance period. PEO Enterprise’s AID team pulled together a requirements document, and the SBIR topic “AI-Enabled ARP, Select and Monitor” was released in April 2024.  

According to the solicitation, by the end of the Phase II effort, the Army expects contractors to “deliver a basic AI capability to guide ARP development and support software development via contract types and clauses, documentation of past successes and the delivery of an AI-built, request for proposal presentable on SAM.gov.” 

By the time the SBIR solicitation closed in mid-June, it had generated 60 responses from small-business industry vendors — a number that’s “significantly larger than normal,” according to Colleran. 

“We blew the doors off this one,” he said.  

Of those five dozen respondents, the Army awarded three contracts on Sept. 4. All three awardees are expected to deliver usable products to help PEO Enterprise acquisition teams with real-life acquisition requirements packages within the next two years, said Colleran.  

“The value they can unlock if they’re successful is tremendous,” said Colleran. For now, he is meeting with the teams biweekly as they navigate through the discovery phase. Once their minimum viable products are delivered by the end of calendar year 2025, PEO Enterprise will figure out logical places for the solutions to exist.  

The Army’s SBIR office is now actively courting PEO Enterprise to develop a solicitation on a companion topic, said Colleran — an AI evaluation tool. This tool will not replace humans as evaluators but help them ensure all proposals are evaluated with the same lens and to the same standard. He sees a “good practical case” for that. 

Colleran’s director, Aric Sherwood — head of the AID office at PEO Enterprise — agrees that AI is a game-changer for the Army and broader DOD when it comes to acquisition. He also sees potential growth for the PEO’s SBIR involvement. 

“We saw an opportunity to jump in where we previously weren’t a participant,” said Sherwood, adding that there still are “lots of nuances to learn.”  

Given industry’s “pretty overwhelming response” to PEO Enterprise’s first-ever SBIR solicitation, Sherwood said he can’t rule out creating a dedicated SBIR office in the future.