Michael W. Derrios
The Potential for Cognitive Procurement in the Federal Space
I've been doing a lot of reading lately on the potential for cognitive procurement in the commercial supply chain and it's made me wonder whether there's any applicability for it in the Federal space. In general, cognitive procurement is a method which uses disruptive technologies to aid in the management of an organization's procurement function. Self-learning technology can process data in smart ways to assist with buying goods and services, and with the procurement line of business already being a very data-rich environment it's a prime candidate for exploring cognitive capabilities. Commercial supply chain managers have access to ample data related to raw materials, goods, transportation, delivery and suppliers that can be leveraged through predictive analytic tools to forecast optimal procurement strategies. This type of data mining can help to improve transactions by speeding up decision making and creating more value through return on investment. But is there any applicability when it comes to the Federal space?
Through Big Data spend analysis, algorithms can help to continuously re-evaluate suppliers and determine which ones have capabilities that best align to a customer's preferences and eliminate underperforming companies from their roster of suppliers. But in a world where competition is king, and required by law, could Federal agencies use Artificial Intelligence (AI) in the same way that private sector companies are starting to do to identify the right sources of supply and services? Well, I say why not? I believe that Federal agencies need to be more focused on harnessing the spend data that we produce routinely through the execution of contracts. We collect data through the Contractor Performance Assessment Reporting System (CPARS) in order to determine whether or not we should award contracts to firms based on their record of past performance, so why not use it to tailor our acquisition strategies upfront?
I'm certainly not suggesting that we stop competing contracts - quite the opposite. We should be using the data that we produce to better inform our market research efforts to promote the right kind of competition. Our talented Contracting professionals are accustomed to checking CPARS after proposals are in but we should be checking the database early on, during the pre-solicitation phase, to see if the companies we think might propose are going to have the kind of track record we need for successful performance and then crafting our solicitation approach based on that information. And, if our potential competitive landscape isn't going to lend itself to our needs, perhaps we adjust our solicitation strategy accordingly. What if CPARS was driven by smart AI that could help the Integrated Project Team (IPT) determine the right way to solicit for competition by mining past performance data and then drawing linkages to our preliminary requirements by analyzing draft Statements of Work? Are our requirements too stringent or ambiguous for the market? Is that GWAC we're considering really the right vehicle for the competencies we need? Should we be considering either set-aside or full-and-open based on what the data tells us? These decisions are sometimes not as well-informed as they need to be and I think cognitive procurement tools could help. Imagine a world where we could create "competition scenarios" and maximize our opportunities for the best outcomes through data-driven analytics that help us tweak our requirements or solicitation methodologies.
Another area where cognitive technologies could help the Federal procurement function is through streamlined processes, such as contract formulation. There are plenty of contract writing softwares on the market with clause logic built in but the problem is that it is template-based and logic-tree driven, centered around a static rule set, which doesn't account for anomalies. And as we all know, Federal contracts are dynamic and one size does not always fit all, for example, when it comes to selecting terms and conditions. The use of AI could help Contracting professionals write smarter contracts that take into consideration all sections of the Uniform Contract Format, on an individual transaction basis, by cross-referencing key elements, flagging potential disconnects, and then delivering a menu of options for terms and conditions or suggested revisions for other language in the contract. It would be like having another set of eyes to help optimize the contract before it's issued.
Cognitive technologies could also help to accelerate contract negotiations on the Federal side. Our counterparts on the commercial side often have the ability to automate their negotiation positions in real time while Contracting professionals in the government are still running Excel drills through manual data input. What if expected price points on materials, volume discounts, labor escalations, profit percentages and overall negotiation positions could all be extrapolated out using machine-learning algorithms? Think about how often a prolonged contract negotiation has impacted mission operations. If there were cognitive technology tools available to augment the human intelligence from our Contracting professionals we could reach contract award milestones faster across Federal agencies.
Here's one more example that might be a little too sci-fi for traditionalists but if you really think about it the upside could be huge. What if there was a cognitive technology based on predictive modeling that could help IPTs determine whether companies will successfully perform on Federal contracts? This is different than using AI to determine if a contract strategy is going to yield the right competitive landscape. I'm talking about literally trying to predict how well we think the eventual successful offeror - whoever that ends up being - is going to do 12 months, 24 months and 36 months into the period of performance based on unique predictor variables. This could be especially helpful on long-term contracts for the development of major systems that involve fluctuating market conditions driven by economic or political factors.
For example, if the Federal government could feed a cognitive technology with realistic information about the expected availability of government-furnished information or property, typical review cycle times on contract deliverables, anticipated constructive changes based on pending legislation, potential interest rate or inflation hikes, or the likelihood of obtaining out-year funding from Congress etc., and couple that with recent and relevant past performance information from expected competitors, it could prove to be a powerful tool for both pre-award and post-award environments. Done the right way, predictive modeling could support source selection teams and bolster contract administration after award.
We should never try to remove the human element from either the Federal or commercial procurement functions but I do think cognitive technology has a place and can deliver value through increased efficiency, effectiveness and cost-savings. And I truly hope that the Federal government can catch up to the private sector when it comes to employing cognitive technologies because mission outcomes and taxpayers could both benefit tremendously.