National Ediscovery Leadership Institute - NELI

🎙️ NeLI Pod Season 2, Ep. 13 is live.

Martha Louks of McDermott Will & Schulte has worked through paper, email, mobile data, TAR, and now generative AI. She's seen every major shift in discovery — and she'll be the first to tell you the fundamentals haven't changed.

Martha joins Daniel Gold and Brandon Mack for a conversation that cuts straight through the hype:

→ Why discovery is now about data, not documents — and why Slack, texts, and chatbot conversations break every traditional workflow
→ GenAI is still TAR: "The same rigor is used in validating results from these new applications"
→ Why AI helps with volume — but privilege, QC, and substantive analysis still need lawyers
→ The cost reality: "Even if we review less as a percentage of the total, that doesn't mean the total amount of stuff that has to get looked at is actually less"
→ Her advice for the next generation: be flexible, process-minded, curious — and skeptical

One of the most grounded, practical conversations we've had on the show.

SPOTIFY: https://open.spotify.com/episode/7HKesHHbFTvVMejgn50qBZ
YOUTUBE: https://youtu.be/l_yZ8pk0_-4
APPLE: https://podcasts.apple.com/us/podcast/neli-pod/id1750535625?i=1000779112847

All season, our guests told us where they thought the law was heading on AI. Then the courts started drawing the lines.

Our new episode, a mid-year recap, is different: no guest chair, just hosts Daniel Gold, Esq. and Brandon Mack putting ten Season 2 guests in direct conversation with the federal case law that landed in the first half of 2026.

Across five topics, a striking pattern emerged. What our judges, scholars, and forensic experts predicted on the show has started showing up almost word for word in published decisions.

What you'll hear:

🔵 Why the federal rules held as the anchor, and how a Southern District of New York opinion echoed Judge Young Kim's exact point

🔵 The emerging split on whether a lawyer's prompts are protected work product, and where courts are drawing the line

🔵 Why validation discipline from the TAR era still governs Gen AI review, and what a Kansas federal court said about where your data lives

🔵 The deepfake and authentication problem the rules of evidence are years behind on

🔵 The human question: billing, access to justice, and what stays authentically ours when the machine does the first draft

A thought-provoking listen for judges, litigators, and anyone watching AI reshape the practice of law!

YOUTUBE
https://www.youtube.com/watch?v=wiI3aoovbrA

SPOTIFY
https://open.spotify.com/episode/6Brs3voWyjDp7ElmktyWts?si=c5c8e47dc7c44c28

APPLE
https://podcasts.apple.com/us/podcast/neli-pod/id1750535625?i=1000777061080

The profession is racing to adopt Gen AI for document review as though it were an entirely new species of technology.

On the NeLI Pod, Dr. Jeremy Pickens of Elevate gave Daniel Gold, Esq. and Brandon Mack a case worth pausing on. It is not new.

Gen AI review is TAR 1.0 in a different form.

His reasoning is hard to set aside. TAR was never a predictive model. It is a process. You learn from a set of documents, build a representation of the matter, then extrapolate across the collection.

Replace the support vector machine with a prompt and you have described Gen AI review exactly. The human authors the model now instead of the machine. Everything else is the same.

That is where it turned serious. Pickens pressed Daniel on a common vendor workflow: read documents, draft a prompt, test it on those same documents, refine, repeat, then run it across the full population.

The problem? You are training on your test set. You think you have captured the whole matter. You may have only overfit to the few documents you happened to read.

His framing stayed with the room. We risk becoming surgeons who skip washing their hands because "this is different, I write prompts." The need for a control set has not changed. We have simply forgotten why it mattered.

Brandon drew out the real question. If Gen AI review is a TAR process at its core, why have we set aside a decade of validation discipline the moment the tool started talking back?

The detour through Parmenides and Heraclitus is worth the listen on its own. 🎧

LINKS:

APPLE
https://podcasts.apple.com/us/podcast/neli-pod/id1750535625?i=1000758651291

SPOTIFY
https://open.spotify.com/episode/0kvubd1cxcdQrJYaT5NH0O

YOUTUBE
https://www.youtube.com/watch?v=m7myuq0G7XY

What if the motion to compel became a thing of the past?

On the NeLI Pod, Chief Magistrate Judge Young Kim of the United States Court of Appeals for the Seventh Circuit offered a vision of discovery's future that sounds radical until you consider it carefully.

In his telling, Rule 34 requests stop being lists of document categories and become prompts. Both sides agree on the prompts at the outset, the Gen AI runs, and the parties live with what it produces.

"Can you imagine a world where we don't have motions to compel?" he asked.

The dispute does not disappear so much as shift. Rather than litigating what was produced, the court turns to whether a given prompt is overly broad or disproportionate to the needs of the case.

What lends the idea its weight is the source. Judge Kim is no technology evangelist. The federal rules are his anchor, and his view is that judges apply them rather than rewrite them. So when a jurist this grounded in the rules forecasts a change this significant, it warrants attention.

Our hosts, Daniel Gold, Esq. and Brandon Mack, navigate us through this thought-provoking episode.

📺 YouTube: https://lnkd.in/gc7NzYSC
🍎 Apple: https://lnkd.in/g5cUZyyv
🎧 Spotify: https://lnkd.in/gK2Dxw3Y

There's a database out there tracking every time a lawyer got caught citing a hallucinated case. Hundreds of entries and counting.

On the NeLI Pod, hosts Daniel Gold and Brandon Mack met with Judge Xavier Rodriguez of the Western District of Texas pointed out what nobody else is saying about it: it's measuring the wrong thing.

Nobody is tracking the thousands of times a lawyer used an AI tool responsibly and it worked exactly as intended. So we've got a running scoreboard of failures with no denominator.

And, as the Judge put it, the lawyers turning in briefs full of fake cases are probably the same ones who would've signed off on whatever a paralegal handed them in the old days — we just didn't declare the world was ending back then.

But the part that should spark real debate is his take on the discovery fights coming our way. Rodriguez is openly frustrated with what he calls the "mother-may-I" approach — parties running to a judge for permission to use AI or TAR before a single document has been produced.

His read of the rules is blunt: a producing party produces, and until a deficiency is actually shown, the court has no business refereeing the process. He's watched us nearly kill TAR with pre-production fights, and he'd hate to watch the profession do it again to Gen AI.

And then the line that'll get a reaction from every litigator: "I don't believe in ESI protocols, actually." His reasoning — parties negotiate and sign them before they've ever tested them, then come running back when the protocol turns out to be wildly over- or under-inclusive.

Worth pulling back up. 🎧

LINKS:

SPOTIFY
https://open.spotify.com/episode/7lELYxLrQMa7gOXPO05fpX?si=5482b78f41964a51

APPLE
https://podcasts.apple.com/us/podcast/neli-pod/id1750535625?i=1000750326407

YOUTUBE
https://www.youtube.com/watch?v=uxNxsybgdSc

Plaintiffs' counsel wants your Gen AI prompts. Are they discoverable?

On the NeLI Pod, Robert Keeling of Redgrave didn't hedge: the prompt is work product, and "it's not even a close question." It reflects the thoughts and mental impressions of counsel about the claims and defenses — classic opinion work product, no different from an interview memo of a key witness.

Then he turned it around with a warning for the requesting side: be careful what you wish for. Gen AI tends to favor plaintiffs — more responsive documents, fewer false positives, often faster.

Litigate hard for the prompts, get one or two judges to rule them discoverable, and you'll slow the adoption of the very technology that was helping you.

He was just as direct on privilege logging — long considered one of the hardest tasks to automate. His verdict: the technology is ready to use today, full stop. After two and a half years working the problem, he's seen Gen AI reliably describe documents and justify privilege claims without revealing the underlying advice.

Not a magic button... it takes iteration, but ready.

And his closing "uncomfortable truth" about AI in legal practice is the line that sticks: "It's not magic. It's math." Very complicated math, but math. It rewards the lawyer who invests the time to use it well, and it can't build your case for you.
The deepfake and evidence-authenticity stretch alone is worth the listen.

Daniel Gold and Brandon Mack got a lot out of this one.

Worth revisiting. 🎧

APPLE
https://podcasts.apple.com/gb/podcast/neli-pod-season-2-ep-1-with-robert-keeling/id1750535625?i=1000744131906

YOUTUBE
https://www.youtube.com/watch?v=qX2KzMuEFmQ

Does AI kill the billable hour, or just raise the rate?

Georgetown Law's Professor Jonah Perlin joined Daniel Gold, Esq. and Brandon Mack to talk about how generative AI is reshaping legal practice and the economics of the profession. A legal writing professor and ethics scholar, Perlin brought a historian's eye to a debate that usually runs on hype.

A few arguments worth carrying with you:

🔵 "AI will kill the billable hour" — not so fast. Perlin treats compensation as hours × rate. Everyone fixates on hours going down. But if hours drop and rate rises, compensation holds — and to a client paying for value all along, nothing has changed. A task worth $10,000 is still worth $10,000.

🔵 The real risk isn't efficiency. It's perceived value. If the billable hour ever falls, it may not be because lawyers work faster — it may be because clients decide the work simply isn't as valuable anymore. That's a far more unsettling outcome than getting more efficient.

🔵 What makes it land is that Perlin is neither evangelist nor doomsayer. His through line: history doesn't repeat, it rhymes. The core of good lawyering survives every turn of the wheel.

SPOTIFY
https://open.spotify.com/episode/4CxitRO5mbeN6INRqET59d?si=31B_uPh7QhWto4LZoSsPFg

APPLE
https://podcasts.apple.com/us/podcast/neli-pod/id1750535625?i=1000764397550

In case you missed it — Episode 11 of NeLI Pod Season 2 with Bridget Mary McCormack is one you'll want to carve out time for this week.�

Bridget is the President & CEO of the American Arbitration Association and former Chief Justice of the Michigan Supreme Court. ��She managed to reframe nearly every assumption our industry holds about AI, access to justice, and what dispute resolution should actually look like.

The conversation, led by Daniel Gold, Esq. and Brandon Mack, created some moments that stayed with us:�

🔹 The AAA saw a 50% year-over-year increase in self-represented party filings — and Bridget calls it unambiguously good news.��

🔹 The world's first AI arbitrator is already live — and its core design principle is making sure every party feels heard and understood before any decision is made. ��

🔹 On remote proceedings: "Knowing what we know now, not offering remote options feels pretty user hostile".��

🔹 On the future of legal information: we're moving from scarcity to abundance — and that's a win for the rule of law.�

Well worth the listen! ��

Spotify: https://open.spotify.com/episode/712MHdHYHzt5MbRSLt9pHR?si=K4RJGvHaRP69HoliioU7Ig

Apple:
https://podcasts.apple.com/us/podcast/neli-pod/id1750535625?i=1000769829383

YouTube:
https://www.youtube.com/watch?v=SJcGar5phhE&t=1s

50% year-over-year increase in self-represented party filings at the AAA.

That's not a typo. And according to our latest guest, it's the best news the legal profession has had in decades.

Bridget Mary McCormack — former Chief Justice of the Michigan Supreme Court and current President & CEO of the American Arbitration Association— joins Daniel Gold, Esq. and Brandon Mack for one of the most thought-provoking conversations we've had on the show.

She doesn't pull punches:

→ Why states that went back to mandatory in-person proceedings after COVID made a mistake

→ How the AAA built and deployed the world's first AI arbitrator — a series of agents that verify they've understood each party's claims before any decision is made

→ Her response to the AI hallucination debate: "We've tolerated human hallucinations forever. We now have technology we can require to show its work."

→ Why the legal system was designed for a world where everyone had a lawyer — and why we haven't updated the operating system in 50 years

→ What a shift from legal scarcity to abundance actually means for the rule of law

If you're on the bench, in practice, or working in eDiscovery, this one reframes the conversation entirely.

LINKS BELOW:
Spotify: https://open.spotify.com/episode/712MHdHYHzt5MbRSLt9pHR?si=uOBPbJBISni-mNw1iVl8Ew

Apple:
https://podcasts.apple.com/us/podcast/neli-pod-season-2-ep-11-with-bridget-mccormick/id1750535625?i=1000769829383

YouTube:
https://www.youtube.com/watch?v=SJcGar5phhE

New episode of NeLI Pod is live, and this one is worth your time!

We had the honor of sitting down with Dean Lumen “Lou” Mulligan of University of Missouri-Kansas City School of Law, NeLI’s long-standing institutional partner. This perspective doesn’t sound like any other you’ll hear in the eDiscovery space. Because Dean Mulligan isn’t coming at this from the practitioner, the judicial, or the vendor side. He’s coming at it from the classroom where the hearts and minds of future lawyers are being shaped every day.

When he spoke with NeLI Pod co-hosts Daniel Gold, Esq. and Brandon Mack about his perspective on AI in legal education, it went well past “we’re integrating tools.” He’s asking the harder questions: What happens to the cognitive foundation of a young lawyer if the reasoning work gets offloaded before it’s ever been built? How do law schools track students toward technology literacy that matches their actual practice area, not just the ones that end up in civil litigation? And perhaps most compellingly, how does AI become a tool for expanding access to legal services for the 80 to 90 percent of people in this country whose legal needs go unmet?

Dean Mulligan makes a point about AI that I don’t hear enough: it’s not just an efficiency play for lawyers who already have clients. The real opportunity is equal access to the law. It’s about equity, fairness, and getting legal services into the hands of people who’ve never been able to afford them.

That’s a fundamentally different framing, and it’s one worth sitting with.
UMKC has been a cornerstone partner of NeLI since our beginning. Having Dean Mulligan on the pod felt like a conversation that was long overdue.

Watch & Listen today! Links in bio.

AccessToJustice NeLI