Making sure the government gets fair and reasonable prices for goods and services is a core function of the procurement system. But getting those assurances can be slow and costly for both agencies and vendors, so the military services are turning to automation for help.
The Army, for instance, is looking to supplement its ranks of pricing specialists with algorithms that have the potential to replace some of the most manpower-intensive aspects of reaching those fairness conclusions.
As part of a pilot project now underway, the service is experimenting with bots that can scrape through the bills of materials (BOMs) that make up certain complex vendor proposals and make determinations about whether each of those line items’ proposed prices are realistic.
“We get these giant, 3,000-page BOM proposals, and we have little bots that go out and crawl the web, crawl Defense Logistics Agency systems, and they’re able to come back within minutes,” said Rebecca Weirick, the deputy assistant secretary of the Army for procurement. “They’re doing what took our price analysts three or four months to analyze, get to prices paid, etc.”
Weirick told a government contracting conference hosted by ProPricer that the automation experiment is one way the Army is trying to reduce the time its contracting officers and contract specialists spend on “low value” work that might be better performed by machines, so that they can apply their skills and training toward the kinds of critical thinking that can only be done by people.
Of course, the more data the bots have access to, the more likely they are to reach reliable pricing conclusions. And the Army has made some inroads toward convincing other DoD components to let its bots crawl through their systems.
Key among those is the Pentagon’s Procurement Integrated Enterprise Environment (PIEE), a collection of data systems that includes, among other things, a central repository of historical contracts — and the documentation used to support those contracts.
“The thirst for data can’t be quenched, but we’re starting to be able to see prices paid for BOM items that have national stock numbers or some other identifiable nomenclature,” she said. “The machine will learn over time as we teach it that, for example, this NSN identified as this nomenclature, and then we’ll get prices paid in enough detail that we will be able, we believe, to probably discern 80% of a BOM as long as we have prices paid data across the department.”
The Air Force has similar ambitions, and started a pilot project of its own through the Air Force Research Laboratory. That effort used a GE Aviation-sponsored initiative called Brilliant Pricing, which claims to be able to cut the time it takes to deliver reliable pricing estimates by 85%. Like the Army project, it uses machines to comb through historical prices for individual components.
“We partnered with them against pricing cases where we could apply automated intelligence to pour through literally huge volumes of pricing data and highlight anomalous pricing. Humans can follow up from there,” said Maj. Gen. Cameron Holt, the deputy assistant secretary of the Air Force for contracting.
If the technology pans out — and if the government is able to apply the security controls that would be required to keep companies from seeing their competitors’ sensitive pricing data — Holt said approaches like these could dramatically shorten procurement lead times, because the government and industry would no longer have to haggle over historical details.
“We could actually get a place where incumbents’ written proposals for follow-on efforts become a thing of the past,” Holt said. “I would even look at whether we could do a very abbreviated negotiation that’s based just upon the uncertainty of the future — and not argue over the past or the present — because we would already know, with absolute certainty, what that looks like.”
In that potential environment, Holt argued DoD procurement officials would be able to issue waivers to the provisions of the Truth in Negotiations Act that require contracting officers to gather certified cost and pricing data when they enter into a sole-source contract much more often, because current and accurate data would already be at the government’s fingertips.
“That could disruptively change the timeline and the ability for us to put things on contract very, very quickly, and avoid the dance and the often unhelpful lack of transparency between the government and its industry partners,” he said.
But that sort of “disruption” would need changes that are as much procedural as they are technological.
In one case, the Navy tackled the procedural part without the addition of bots or artificial intelligence, and came away with results that still shortened the procurement timeline.
In its most recent lot of F/A-18 fighter purchases, the Navy decided to streamline the buying process by not requiring Boeing to submit the sort of detailed cost data that would normally accompany a multibillion dollar procurement. The logic, in that instance, is that after 30 years of buying roughly the same airplane, the government has enough of its own data to figure out whether the asking price is reasonable, even if there’s only one source.
“We felt it was truly fair and reasonable, based upon good, historical data,” said Cindy Shaver, the deputy assistant secretary of the Navy for procurement.
That approach was helped by yet another pilot project — a procedural one, not a technological one — that Congress approved as part of the 2019 National Defense Authorization Act.
At DoD’s request, lawmakers authorized the Pentagon to test sole-source procurements where, like in the F-18 example, procurement officials can “tailor” the traditional requirements in the Truth in Negotiations Act.
Under those trimmed-back legal requirements, program officials, with high-level Pentagon approval, are allowed to enter into deals with sole-source vendors based mostly on the government’s own data about historical costs, rather than demanding that vendors furnish – and that program offices sort through – mountains of paperwork to determine whether each bolt and washer was priced accurately.
“I think it’s one of the first big steps in this process of trying to really figure out what makes sense for an individual procurement, based upon what you’re buying and who you’re buying it from,” Shaver said. “The door is open, or at least it cracked for us to demonstrate that we can make the right risk equation. There are going to be circumstances where this absolutely will not make sense, but there are a lot of circumstances where it does.”