When lawyers hear AI, many think of ChatGPT. They think of hallucinations, made-up citations, and sanctions. They think of the attorneys who submitted briefs with fabricated case law because they trusted an open model without validation.
HaystackID has spent years putting safeguards in place so that risk is controlled.
Aaron Pribil and Todd Haley are part of a data services company that has been testing AI in eDiscovery for over five years. They did not rush to market when generative AI captured headlines. They ran proof-of-concept projects. They validated results. They built governance frameworks. Only then did they put AI in front of clients.
The Validation Problem
Ask legal teams about their biggest concern with AI, and the answer comes back consistently: trust. Can they trust the results? Can they defend those results in court?
"The biggest concern is making sure that there's validation around the AI," Haley explained. "People have concerns around hallucinations, which is why we use other methodologies that we've been using for years to ensure that we validate the results."
The validation shows up in the numbers. As AI is applied within HaystackID’s CoreFlex™, results are evaluated using established methods, with performance reaching upwards of 90 percent recall and precision in the high 80s. Those metrics matter because they translate directly to defensibility. When opposing counsel challenges your review methodology, you need documentation showing how you got there.
"There are lawyers that have used AI but didn't validate its results, and that can lead to sanctions and other repercussions," Pribil noted. "Validation is critical."
Closed Systems Change Everything
The distinction between open and closed AI systems matters enormously in regulated environments. Open systems train on public data and can leak information. Closed systems operate only on the data you provide.
“When you're using these large open models, you're drawing from the whole world’s worth of information,” Pribil explained. “For us, AI—especially in the legal and regulatory environment we operate in—is heavily scrutinized. We have to make sure the data is grounded very specifically to the legal issues at hand.”
Haley pushed back on the assumption that all AI carries the same risks. "When you're talking about open AI systems, a lot of what you're talking about is right. But in the closed systems that are designed for purpose, the AI does not have the same challenges. It is very organized, structured, works within a particular LLM, and ensures that the data is working with the data that you're giving it."
HaystackID's systems meet SOC and ISO requirements around security. The AI never touches public data or public information. That architecture enables work with some of the largest banks in the world, institutions operating under layers of regulation.
The Intake Process Nobody Talks About
Speed dominates most AI marketing. HaystackID leads with intake.
Their platform, CoreFlex, functions as a command and control system where clients manage matters, users, and data. But the real value comes from what happens before any AI analysis begins.
"In that intake process, making sure that they're setting up their matters appropriately, that they're getting the appropriate data that needs to come into the ecosystem," Pribil explained. "We actually break down what is the information that we need categorically to help build out AI solutions for them, whether it's for privilege, relevance."
That information flows to HaystackID's Legal Data Intelligence team, subject matter experts who review the inputs, consult with clients, and provide feedback before anything hits the language model.
"Garbage in, garbage out, you have to be careful of," Haley said. "We have to have that type of high-end quality first and foremost, and then get the speed."
One client who had completed AI projects with other industry leaders told HaystackID that the consultative component made a significant difference in their overall review, privilege analysis, and cost savings. The preparation work matters.
Measure Twice, Cut Once
The efficiency gains from AI in eiscovery are real. Research from the 2025 Ediscovery Innovation Report found that 42 percent of AI users save between one to five hours weekly on document review. Extrapolated annually, that represents 260 hours saved per individual, equivalent to 32.5 working days.
HaystackID sees similar results. "We're looking at 60 percent to 70 percent savings in time, cost, et cetera," Haley confirmed. "But I also want to caution individuals that it's not an easy button."
He invoked his father's carpentry wisdom: measure twice, cut once. "One of the challenges I find when watching some of the people using AI right now is they're cutting and then realizing they measured wrong. If we take a few days upfront to organize and plan, the backend benefits are exponentially greater."
That patience runs counter to market pressure. According to the American Bar Association, AI legal document review can automate processes including eDiscovery, document summarization, and drafting. But automation without governance creates new problems.
"If you allow the tool to do your job for you, then you're doing it wrong," Haley emphasized. "It is a tool. It is not a replacement for you."
The Black Box That Is Not
Critics often describe AI as a black box. HaystackID takes a different view.
"There's a mathematician that completely understands the algorithm of how LLM works," Haley explained. "Our job is to take that mathematical precision and understanding and convert it into layman's terms. When we talked earlier about intake, about how we manage the data, how we govern it, how we validate against the results, all of that eliminates that black box problem."
The transparency comes through documentation. Courts increasingly accept technology-assisted review, but defensibility requires clear protocols and validation steps. HaystackID builds that documentation into the workflow rather than treating it as an afterthought.
"At the end of the day, we understand what we put in, we understand what came out, and we know why what came out ended up being there," Haley said.
The Speed of Adoption
What surprised the HaystackID team most was how quickly legal professionals embraced AI after years of slow technology adoption.
"Being in this industry, the overall general premise of technology adoption in our space has been slow," Pribil observed. "It's unbelievable how quickly AI has flipped the script on that. Everybody is rushing to figure out how to institute this new technology."
But the rush creates its own risks. Pribil highlighted the case of a client using ChatGPT for legal advice, only to have those conversations become discoverable in court. The AI interaction was not protected by attorney-client privilege.
"It's those things that people don't think about that with us and our legal hats on are always very mindful of what we're doing with AI," he noted.
Advice for Companies Building AI Systems
HaystackID's guidance for organizations tackling AI initiatives boils down to focus and patience.
"Listen to the market and to your clients," Pribil advised. "You need to serve something that clients are willing to use, and there's a need for."
Haley recounted meeting with a corporate client that had over 40 different AI initiatives underway. His response: pick the top five. Get those right in 2026. Use the lessons learned to tackle the remaining 35.
"We didn't roll it out until we were ensured that we had done the first POC ever done, that we had separate validation methods that we had used, and we were confident with the results before we put it in front of our clients," Haley said.
What Comes Next
HaystackID is focused on further AI integrations and making data ingestion easier through self-service capabilities. But the more interesting direction is what Pribil calls moving left.
"Moving left means the origins of where the data started," he explained. "How can we help our clients mitigate risk, prepare for something that's incoming, and get them from A to Z if something happens legality wise."
The vision positions HaystackID not just as a document review service but as a risk mitigation partner throughout the litigation lifecycle.
"In our argument, quality must come first," Haley concluded. "And then get the speed. The beauty of it is we are pretty good at getting them both."
Enjoying insights from industry leaders? Subscribe to The Winners' Circle podcast on your favorite podcast player and never miss an episode. Listen and subscribe at bintelligence.com/podcast.