The process that school districts follow in putting forward requests for proposals, and that vendors follow in bidding for work, is a winding road of compliance checks, clarifying questions, evaluation, and scoring.
Depending on what’s being purchased, the timeline can take many months, or over a year.
Now, school systems and education companies are both taking tentative steps to bring an increasingly omnipresent technology – artificial intelligence – to procurement, in an effort to streamline the process, and in the case of vendors, make their bids more competitive.
Some school districts have begun experimenting with AI as a way to make sure their RFPs include the language they need to secure the products they want and protect their school systems’ interests. Vendors, for their part, have started using the technology to automate responses to K-12 solicitations and ensure they’re meeting every requirement.
Key Takeaways
- School districts are using AI to draft clearer, more comprehensive RFPs. The technology assists in both editing and researching when putting together a solicitation.
- Education companies see benefits in AI helping them understand the requirements of RFPs, and responding to them in a timely manner.
- Having a human check on AI-generated content is paramount. District officials say it’s obvious when AI-powered RFPs are submitted hurriedly and without human review.
- Public entities, including school districts, are increasingly asking for vendors to disclose if and where AI is used in their bids, in a push toward transparency. Some say companies can be disqualified if they are not forthright.
School district officials told EdWeek Market Brief that they’re already seeing more responses to RFPs come in from vendors that use AI. Those administrators said they generally don’t have objections – as long as companies maintain transparency about how they relied on the technology (which is a legal requirement in some cases). The districts said they also want to know that the vendors’ bids went through human review, and that the bidders have authenticated their ability to fulfill the request.
The use of artificial intelligence in the drafting of RFPs and responses to them, however, does raise some concerns that the technology could open doors to bias, inaccuracies, or generally low-quality responses as companies tap into AI to submit as many bids as possible.
“We’re all compelled to do our work as efficiently as we possibly can, but it comes down to, are we being safe and ethical?” said Brianne Ford, president of the Education Technology Joint Powers Authority, which negotiates and secures competitive technology contracts on behalf of its member school districts across California.
“In the end, whether it’s created by people or it’s created by AI, the organization [must be] accountable for the result and must stand behind the language they chose and the work they did,” she added.
EdWeek Market Brief spoke to education company officials and those leading district procurements to discuss how AI is being used in RFPs and the benefits and limitations K-12 administrators and companies see in its early applications.
More Relevant, Up-to-Date RFPs
School systems are only beginning to explore the potential for AI to cut the amount of time they devote to crafting RFPs, those who work on procurement say.
Equalis Group, a cooperative purchasing organization that serves about 1,300 K-12 school districts nationwide, has trained its own AI model to help schools and other public agencies develop solicitations.
Each step of the process allows for manual editing of results, or a reframing of prompts to get desired output, whether it’s drafting a description of the product they’re searching for, or refining the questions the school system is asking for in its evaluation criteria.
The AI model is especially helpful in cases where a district doesn’t know where to begin in crafting its proposal, said David Akers, executive vice president of the group.
If top district procurement leaders are putting out bids for a new type of technology for the first time, for example, they probably don’t have the expertise necessary to know all of the components they’re going to need, he said.
“How do we put together the solicitation?” Akers said. “What questions should we be asking, how are we going to evaluate responses?
You’re going to get better output in your solicitations if you’re effectively leveraging AI, which means that you should be getting better proposals, which means you should be getting better solutions.
David Akers, Executive Vice President, Equalis Group
Those fundamental questions also apply to items that schools procure once every few years, Akers said. Many procurement officials will just reuse a previous RFP, updating dates and minor details, or they’ll ask around to see if peers in other school systems have a template.
But in the years since they last called for bids, the capabilities of the product and the needs of the district may have changed significantly. AI can assist in sourcing updated information to pull together an RFP that is timely and relevant.
Artificial intelligence can help school systems in “developing better solicitations to get what you really need,” Akers said. “You’re going to get better output in your solicitations if you’re effectively leveraging AI, which means that you should be getting better proposals, which means you should be getting better solutions.”
The Jordan School District in Utah, which serves 56,000 students, is one specific school system that has begun using AI to create stronger RFPs and to cut down the drafting time.
Putting together a list of comprehensive criteria questions for an RFP used to take hours for Tonya Hodges, a senior buyer for the district’s purchasing department. She would scour the internet to see what other states or school districts were listing as criteria to compare to her own ideas to see what she was missing.
Now, she can input a detailed prompt into a chatbot, which can quickly compile a list of criteria and a description for the scope of work – all while eliminating redundancies in the content and making sure the text reads clearly. She and her colleagues will then review the document.
“This is still a new space, so we do a bit of double checking to make sure that the information coming back is not a hallucination, but it’s been great that I don’t have to sit and search other districts,” Hodges said.
Using AI to Submit a Bid
Many vendors are already using AI to help them craft responses to school district RFPs.
Requests for proposals are often lengthy documents with different layers of questions and requirements. The first step a company must take, before preparing a response, is to understand what’s being asked of it.
Company officials say artificial intelligence can help them respond to more RFPs, and more quickly, by allowing them to break down the task at hand and make sure their proposals have met all requirements listed on the rubric.
“The biggest thing with RFPs is to really understand what the district is looking for,” said Jack Friedman, founder and CEO of Study Smart Tutors, which provides in-school and after-school tutoring and intervention programming.
Artificial intelligence can also help a vendor collect all relevant, disparate information it needs to piece a proposal together quickly, in order to meet the deadlines districts put on solicitations, Friedman said.
Some companies “don’t have their information collected, and they say, ‘If you give me three months, I can get this to you,’” Friedman said. “But if they need this information by next [week], and you don’t have your processes down and your information in the right place, there’s no way you can respond, even if [the RFP is] a great fit for you.”
[If] you don’t have your processes down and your information in the right place, there’s no way you can respond, even if [the RFP is] a great fit for you.
Jack Friedman, Founder & CEO, Study Smart Tutors
About 60 percent of school district RFPs ask for the same information, according to Rye Consulting, an education-focused consulting firm. Those boilerplate requirements include things like the history of the company, research, qualifications, and experience.
“This is where the opportunity for AI can play a role,” said Rayna Glumac, managing principal consultant for the firm. “Building a repository of language and making it easy for companies to access can help efficiently build out templates and [give them] a leg up on the quick turnaround time required by RFPs.”
Vendors are also using AI to improve the nature of the language they use in responding to RFPs, said Brent Mital, senior research analyst at Deltek. a provider of software and solutions for project-based businesses. Deltek uses AI to help vendors find bids through a database it manages of RFPs from state and local governments, including school districts.
An AI model can help companies refine the language of their bids or rearrange the structure of a proposal for clarity, Mital said.
“Some proposals are science-oriented or tech-oriented,” he said. “That’s where using AI can help clarify points that may have taken more time in the past because you’re talking about something really technical.”
When AI Goes Unchecked
Though some companies worry that districts will not approve of their use of AI in responding to RFPs, a number of K-12 procurement administrators said that keeping the technology out of the process is probably unrealistic.
“I don’t have any opposition to vendors using AI to respond, as long as there is that [human] review,” said Ford, president of the Education Technology Joint Powers Authority. “Because it does save time and makes it a lot less difficult to respond to the RFP, and we get more direct and substantive responses to each requirement.”
But when companies use AI to crank out RFPs without a person taking the time to guide their submissions, the difference in quality is obvious, said Ford, who also serves as the assistant superintendent for information technology at the 38,000-student Irvine Unified School District in California.
Proposals driven entirely or mostly by AI are often riddled with inaccurate information or details about the wrong district, or they don’t appear to have been proofread, she said.
They’re also more likely to have a “gross misinterpretation of the question,” or they could look like a lot of filler information without a substantive response to the requirements.
Even when proposals that aren’t assisted by AI come in, “you can tell when [they] are rushed or when the sales and product teams are on two very different pages,” Ford said. Bids that are submitted using AI tend to have “really generalized responses.”
Ford likens it to a student trying to write an essay based on a book they didn’t read.
“It’s very high-level – a lot of language that says nothing,” she said. “They’re responding to the criteria without any real substance because the AI product doesn’t know their tool well enough.”
Another giveaway is that AI cannot interpret context. Within the areas of technology and special education, for example, there can be the same acronym – with two different definitions depending on the context. AI – without human review – can mix up the meanings, leading to a response that is “contextually bizarre,” Ford said.
A Push for Transparency
Some public entities, including school districts, are including language in their RFPs requiring vendors – if they’re using AI to write or submit a proposal – to divulge explicitly where they’re using it, and to what extent.
The state of California is one example of a government buyer making that requirement. Some of its state government RFPs include a multi-page form with specific sections to be filled out – requiring in-depth detail about AI use and the portions of the proposal it was used for.
This form is standard on most, if not all, of California’s solicitations in recent months, said Mital, of Deltek.
In some cases, buyers make it clear that if it’s discovered that AI was used in the proposal but not disclosed, that could disqualify the company in the solicitation process, he added.
An RFP issued last year by the Hampton City Schools in Virginia for internal broadband services says that vendors can be disqualified if their proposals include ideas for service “proposed by an artificial intelligence system that does not take into consideration the specific needs” of that district.
“The more [information] you can disclose, the better,” Mital said. “I would be upfront about it.”
You could have the best-written RFP, and you could provide the best service for what they’re asking for, but if you are not transparent, that could completely disqualify you.
Brent Mital, Senior Research Analyst, Deltek
Even in RFPs that don’t specifically ask for details on AI use, it’s a good idea for companies to divulge that information anyway just to be clear. The information they disclose could include whether they used AI in a specific section or to help with big-picture support, for things like overall outline development, he said.
“You could have the best-written RFP, and you could provide the best service for what they’re asking for, but if you are not transparent, that could completely disqualify you,” Mital said. “It’s not worth the risk to hide anything.”
And it’s critical to make time for personal review of any proposal that relied on AI, said Hodges, the senior buyer from Utah’s Jordan School District.
“My concern when [bids] are generated by AI is that it’s done in a way that makes it sound like they’re meeting all the specifications, or exceeding them, when in reality, we don’t know if it’s really true,” Hodges said.
“So don’t just throw everything into a GPT and send us whatever it spits out,” she added. “We can tell when [vendors] spend a lot of time on a response and when they haven’t. If this is how they’ve written their RFP response, chances are this is how they’re going to handle the project.”