[Podcast] HaystackID® in the EDRM Illumination Zone: Gabe Landau

The shortcut that looked like innovation turned out to be a liability wearing a better interface. That’s how Gabe Landau describes a technology he watched enter the breach-notification market years ago, a tool that promised to skip the hardest part of incident response entirely: identifying whose personal information had been exposed and building the list of people a company legally had to notify. Press one button, and the software will identify and extract PII, then generate a finished notification list. No manual review, no bottleneck, no handwringing. It gained traction fast, right up until a sophisticated plaintiff attorney asked a question the easy button couldn’t answer: who’s attesting to this list, and how did you arrive at it?

Landau, Senior Vice President of eDiscovery and Incident Response Solutions at HaystackID, told that story not as a cautionary tale about one bad product, but as a lens for how he evaluates every new technology that crosses his desk, including the current wave of GenAI tools reshaping his industry. He recently returned to EDRM’s Illumination Zone podcast, joining hosts Mary Mack and Holley Robinson for a conversation about defensibility, trust, and what separates genuine progress from a faster-looking version of the same risk.

Getting Ahead of the Data, Not Just the Deadline

Incident response engagements typically run into the same wall: nobody fully understands their own data until someone actually looks at it. Custodial interviews only go so far. Assumptions about what’s sensitive, where it lives, and who touched it tend to collapse the moment real analysis begins.

GenAI has started closing that gap earlier than ever, and Landau considers this one of the more consequential shifts he’s witnessed in his career. Historically, he explained, technology got deployed only after a matter moved through traditional processing and search, then applied downstream for review and tagging. Now that analysis has moved to the front of the process, into early case assessment, classification, and PII detection, reshaping both the cost of an engagement and the accuracy of the strategy built around it.

“The biggest force multiplier that I’m seeing for sure is how quickly clients can begin understanding their data at the onset of a matter,” Landau said.

He pointed to one matter, six million documents in early case assessment, where that population dropped 90% almost immediately, using conversational tools that let a case team interrogate their own data directly. It wasn’t just a spend that got informed, Landau said. It was accuracy, changing what the team understood about the case before major decisions were locked in.

Landau is careful, though, not to oversell where the technology stands.

“There’s still more promise than practice, this idea that AI is going to take over,” he said. “I don’t think that’s going to happen wholesale. I think it’s going to augment how we work.”

The tools can accelerate the work, in his view, but they don’t supply professional judgment on their own. That’s a permanent part of the workflow, not a placeholder waiting to be automated.

The Forensic-First Instinct

Capability is rarely where Landau starts when a client is weighing a GenAI-based workflow for the first time. Defensibility is, and HaystackID’s track record on that front predates any legal precedent that might have forced the issue.

“I know this won’t be evergreen now that I’m going to say it,” Landau said, before pointing to Schulte v. LinkedIn Corp., a recent case that, in his view, treated GenAI defensibility as equivalent to the validation standards long applied to technology-assisted review: statistical sampling, elusion testing, documented methodology.

HaystackID built its GenAI deployment around those same principles well before that case law existed, a byproduct of the firm’s roots as a forensics-first company that never had the option of treating validation as optional.

“We mitigate risk and take a defensible posture in every facet of what we do,” Landau said. “We always started as a forensic-first company. And so, before we even deployed enterprise-wide our GenAI solutions, we had defensible measures that we took because you have to in this industry.”

That discipline gets tested the moment a regulator or opposing counsel starts asking hard questions. In Landau’s experience, a shortcut that can’t explain itself eventually meets someone whose job is to ask it to.

“In no world are you going to be able to say that the technology said it was this and that’s what we did,” Landau said. “That’s just not going to fly. You’ve been through these enough times to know that any sophisticated counsel, federal agency, or state agency is going to interrogate that.”

Risk tolerance, in this view, is entirely bespoke to the case and the client. Sometimes that means slowing down rather than running a full GenAI review; it doesn’t mean avoiding the technology altogether. Privilege logging and quality control, for instance, are places where GenAI can create real efficiency without introducing the kind of exposure a fully automated notification list would.

The Four Questions Incidents Eventually Answer

Speed and accuracy tend to get treated as competing priorities in incident response, when in practice, they’re the same problem viewed from different angles. Engagement, regardless of the client’s size or sector, collapses into the same four questions:

• What happened?
• What systems and data were affected?
• Whose information was involved?
• What obligations does the organization have as a result?

“Incident response is a good example of why speed and accuracy have to work together,” Landau said.

Answering those four questions under real-world conditions-encrypted files, incomplete logs, data scattered across dozens of systems-takes considerable coordination. Once data has been preserved, collected, processed, and normalized, GenAI can identify PII, PHI, financial records, and other sensitive categories far faster than manual review alone. It can summarize, prioritize, and extract information that used to take teams days to surface, narrowing the gap between collection and a genuine understanding of what’s at risk.

But no single tool solves the entire lifecycle on its own. The harder problem is connecting forensics, data analytics, discovery, document review, and notification analysis into one coherent process, especially when the underlying data resists that kind of cohesion. That’s the piece still catching up: not identifying sensitive information but stitching every piece of that identification into a single, defensible narrative.

“The goal is not to simply reach an answer faster,” Landau said. “It’s to reach a reliable answer faster.”

The Ledger Nobody Can Shortcut

Underneath the technology and process sits something harder to quantify: trust. Landau returns to this idea more than once, and it’s worth sitting with why he treats it as central rather than incidental. It’s also, not coincidentally, why he was recently named a finalist for Relativity’s Customer Experience Innovation Award, a category he says means more to him than most because it recognizes relationships rather than product or workflow.

“Customer experience begins at the first interaction with somebody well before any kind of sale happens, or any sort of transaction happens,” Landau said. “It’s your interactions, it’s your understanding, your comprehension, your patience, your empathy, your ability to understand or try to see and really listen and feel what somebody is going through.”

Incident response engagements don’t just involve data. They involve people’s careers, companies’ reputations, and sometimes the safety of individuals whose personal information has been exposed.

“Trust is really, really important for me and for everyone at HaystackID. Someone else’s reputation, someone else’s lives, and the lives of people that work at these companies potentially are at risk at any one of these cases,” Landau said.

That trust doesn’t come from a single successful engagement. It gets built through how a firm handles uncertainty, how transparently it communicates when results don’t land where expected, and how willing it is to walk a client through the granular mechanics of validation, statistical sampling, elusion testing, unresponsive-document handling, rather than gesturing vaguely at a QC process and hoping it satisfies the room. Landau describes GenAI output the same way: never accepted at full value without human oversight, because the technology’s speed only matters if a person remains accountable for what it produces.

Moving Left of the Breach

Looking ahead, Landau doesn’t expect the next major shift in eDiscovery to center on faster review or smarter classification models, at least not primarily. He expects the industry to keep moving earlier in the timeline, away from matter-specific work and toward continuous enterprise data intelligence.

“I do believe that we’re going to continue to move left and start talking less about matter-specific activity and more towards enterprise data intelligence,” he said.

The same technology and expertise that helps organizations understand their data during a breach, he argued, can just as easily support information governance, internal investigations, regulatory readiness, and ongoing risk management, long before a lawsuit or incident forces the issue.

That shift carries its own complications. As AI-generated and manipulated content becomes more common, Landau expects authenticity and validation to become their own discipline, one that organizations will need well before any breach makes it urgent. He’s candid that the industry has talked about better data hygiene for years without much to show for it; cyber incidents haven’t moved that needle the way many predicted. Whether enterprise data intelligence finally forces the issue, or whether it takes something further out on the horizon, remains an open question.

As Landau put it, only half joking, “if this doesn’t, then quantum computing should.”

What’s clear from the conversation is that Landau doesn’t measure progress in eDiscovery by how much faster a tool can process a document set. He measures it by how confidently a team can stand behind what that tool produced, months or years later, in front of someone whose job is to find the crack in the story.

More About Gabe Landau

As the Senior Vice President of eDiscovery and Incident Response Solutions at HaystackID, Landau’s professional focus lies at the intersection of law and technology, where he brings his expertise to deliver practical, defensible, and cost-efficient approaches for clients’ eDiscovery and Cyber Discovery requirements. Prior to his work in the legal sector, Landau garnered experience as an Emmy Award-Winning Producer and Director in the realm of reality TV. His achievements extended to receiving accolades such as the Telly, Davey, and Webby Awards. Drawing from his diverse background, Landau seamlessly integrated his skills into the legal field, enhancing client experiences in the legal sector through the application of advanced technology. He graduated Cum Laude from Ithaca University with a Bachelor of Science degree.