Fulfilling government contracts requires procurement specialists to spend weeks methodically combing through databases and documents, while meeting unforgiving deadlines.

Today, this situation is rapidly changing as artificial intelligence transforms this traditionally manual process, offering new tools that can reduce proposal writing time by up to 70%.

AI adoption is reshaping how companies build their government offerings, said Joe Schurman, partner and principal at PricewaterhouseCoopers (PwC), which leads AI services for the aerospace and defense sector. of the company and co-leads the company’s U.S. space program.

From identifying opportunities to drafting complex technical responses, he says, AI streamlines almost every aspect of the public procurement process.

PwC, in April 2023, announced a billion-dollar investment to expand and scale its global AI offerings. She has created specific AI solutions for aerospace and defense, including AI for Flight Science and Engineering Design and PartsGPT. “But AI for proposal writing is the most commonly requested application in our practice,” Schurman told SpaceNews.

These are not typical chatbots. The next generation of AI tools are purpose-built to navigate the complex world of public procurement, where security protocols are strict and the stakes are high because a single proposal can make or break a company’s financial year. business.

Beyond basic automation

Technology goes far beyond simple document processing. Modern AI systems can analyze large databases of government solicitations on SAM.gov (the official U.S. government procurement portal) to identify contracts that match a company’s capabilities. They can break down complex requests for proposals (RFPs) into manageable tasks and even help write proposals that highlight a company’s strengths, while ensuring compliance with strict government regulations.

For niche entrepreneurs in the space industry, for example, AI’s ability to precisely target relevant opportunities eliminates countless hours of manual searching through federal databases, Schurman says. “It can be difficult for a commercial space company to determine which Space Force contract to bid on,” he explained. “This technology helps them avoid wasting time searching for unsuitable opportunities and focus specifically on their sweet spot.”

Transform the proposal process

The traditional process of responding to government tenders has long been resource intensive. Teams of writers, subject matter experts, and compliance officers typically spend weeks analyzing hundreds of pages of requirements, ensuring that every specification is met and every question is answered. Some RFPs can have deadlines as short as 30 days, requiring companies to submit complete proposals quickly.

“Now they’ve reduced the time it takes to take out these remedial tasks by at least 60 to 70 percent and then leverage these tools as accelerators to help them respond,” Schurman said. AI systems can automatically summarize entire RFPs, break them down into actionable pieces, and automate proposal responses.

One particularly powerful application is the ability to process large government regulatory documents. In discussions with a satellite vendor, for example, PwC provided a proof of concept leveraging generative AI to better understand NASA certification requirements. Schurman said the AI ​​not only broke down the gigantic set of documents, but also helped prioritize which requirements to complete first.

The human element remains essential

Despite the efficiency gains, industry experts emphasize that AI is not a replacement for human expertise. “Our position as a company is that AI is not a replacement tool. It’s an acceleration tool,” Schurman said. “You want individuals and experts in your company with this level of intelligence and passion for the project to infuse their human creativity into a response to a proposal, but leverage generative AI services to remove latent human steps of the process. »

This view is shared by Sean Williams, co-founder and CEO of AutogenAI, a London-based startup that is expanding its AI-powered proposal writing services in the United States.

“Humans understand the logical context and nuance of an argument much better,” Williams said in an interview. “AI generates ideas, but humans structure the response. »

Security Compliance

One of the biggest challenges in integrating AI into public procurement has been ensuring data security compliance. Commercial AI tools like ChatGPT are not allowed on secure government cloud environments required for defense contractors. All AI tools must therefore be built on authorized platforms.

Schurman said PwC’s solution leverages Microsoft Azure OpenAI, one of the platforms approved to manage DoD proposal data and which adheres to the strict security standards of the Pentagon and the International Traffic in Arms Regulators (ITAR). ).

Amazon Web Services, Google Cloud, and Oracle Cloud have also recently been authorized to host sensitive DoD workloads.

Cloud systems must meet DoD security requirements, called “impact levels,” which range from Level 2 for public information to Level 5 for highly sensitive critical data. “These capabilities are finally working at DoD Impact Levels 4 and 5, and even at the classified level, it’s awesome,” Schurman said. “You’re going to see huge advances.”

Demand for AI-powered development and proposal writing tools has attracted some venture capital investments. AutogenAI’s recent $39.5 million Series B funding round, co-led by Salesforce Ventures and Spark Capital, brings its total venture investments to $65.3 million.

It’s becoming clear that this technology is poised to transform the lengthy and highly complex RFP process, said Nowi Kallen, managing director of AI at Salesforce Ventures.

The military’s point of view

The military itself is cautiously adopting AI in its procurement processes. Col. Heather Bogstie, who oversees space systems integration within the Space Force’s Space Systems Command, highlights the Department of the Air Force’s recent deployment of NIPR-GPT, a generative AI program intended to be used on the Unclassified Internet Protocol Router (NIPR). secure military network.

“This allows for greater adoption and advocacy for the use of AI, and removes a lot of the stigma,” Bogstie said at the recent MilSat symposium in Mountain View, California.

The military ultimately aims to use AI to evaluate industrial proposals, which is currently an entirely manual process. “We would like to see more of them used in the acquisition lifecycle to help us with source selection, technical evaluations, proposals, RFPs, etc., so we can speed up our daily tasks.” » added Bogstie.

AI learns government lexicon

Some businesses are already seeing the benefits of adopting AI. Aalyria, a space communications technology company spun off from Google parent Alphabet, uses large language models to adapt proposals to the technical language of certain agencies.

“In our industry, there are often 10 different jargon words that mean the same thing,” said Brian Barritt, Aalyria’s chief technology officer. “We’re finding that AI is really helpful in supporting the response to proposals by training it on previous writings and having it respond in the correct language expected by the government.”

The company has developed a system that can be trained on already written proposals and agency-specific terminology, ensuring that new submissions match the expected language and format. This attention to linguistic detail can be crucial in public procurement, where precise terminology can mean the difference between winning or losing a tender.

Future prospects and challenges

There is still no hard data on whether AI-assisted proposals will win more contracts, AutogenAI’s Williams said. However, early feedback suggests the technology is delivering results. “It will probably take us another 12 months before the procurements are finalized, and we start to have a solid evidence base,” he said. “But the first signs, the first anecdotal data, are very positive.”

Technology continues to evolve rapidly. PwC’s Schurman, however, emphasizes the continued importance of human oversight.

AI hallucinations are common, he explained. “There must therefore always be a human to manage the AI, in order to achieve the required levels of quality and compliance. The relationship between AI and humans, particularly in aerospace and defense, is delicate. The balance of responsibilities will shift as all relationships do. What we want to avoid is a one-sided situation.

As more companies adopt these tools and the technology evolves, he predicts the public procurement landscape could look very different in the future. AI can speed up and improve the proposal process, but human judgment remains essential in public procurement.

This article first appeared in the November 2024 issue of SpaceNews magazine.

Related