## Detailed walkthrough
### [00:04](https://www.youtube.com/watch?v=yQEE_bXt5DE&t=4s) Introduction and Initial AI Priorities
John Buyers, a partner at the international law firm CMS, opens the panel by framing his role as "poacher turned gamekeeper" — an intended beneficiary of legal AI who also advises clients on its implementation and governance. He asks the panelists to state their single most critical AI priority for the year. Hamid Kohan emphasizes that his focus on the plaintiff firm market brings distinct operational challenges.
```
+-----------------------------------------------------------------+
| LEGAL AI ROADMAP |
+-----------------------------------------------------------------+
| Hamid Kohan (Practice AI) | Operational AI (Intake to Docs) |
| Blake Rooney (Wells Fargo) | Enterprise Efficiency & Vendors |
| Emily Chapman (DragonGC) | Verifiable SEC AI Workflows |
| Eric Tan (Goodwin) | Reimagining Client Experience |
+-----------------------------------------------------------------+
```
Blake Rooney highlights efficiency and vendor management as his dual focus, while Emily Drazan Chapman targets the gap between technical capability and attorney trust. Eric Tan describes his goal of modernizing the law firm client experience, observing that entering legal tech after building law-enforcement drones felt like stepping back 20 to 30 years.
- **Hamid Kohan** (Founder/CEO, Practice AI): Focused on operational AI from client intake to document collections and drafting for plaintiff firms.
- **Blake Rooney** (Managing Director, Wells Fargo): Managing AI adoption across a massive enterprise after 20 years at Husch Blackwell.
- **Emily Drazan Chapman** (Legal AI Architect, DragonGC): Developing auditable disclosure workflows for SEC compliance.
- **Eric Tan** (Chief Digital & Technology Officer, Goodwin): Overseeing digital transformation across a 3,000-person law firm.
---
[03:00](https://www.youtube.com/watch?v=yQEE_bXt5DE&t=180s) Specific AI Initiatives in Action
The panel transitions to concrete deployment stories. Blake describes how Wells Fargo leverages its internal IT scale to build custom agents natively alongside commercial legal platforms like Harvey. Emily details how DragonGC automates the labor-intensive research behind nominating committee decisions and SEC filings.
Eric introduces Goodwin's internal innovation strategy, the "Goodwin Builders" program. By pairing computer science-trained attorneys with software engineers, the firm creates targeted tools built by lawyers for lawyers.
- **Wells Fargo**: Engaging line attorneys directly to co-design internal task automation agents.
- **DragonGC**: Automating exchange and SEC requirement lookups to compress hours of director skills research into instant comparisons.
- **Goodwin Builders**: Recruiting hybrid lawyer-developers (such as Harvard Law graduates with CS degrees) to build bespoke M&A and venture financing tools.
---
[06:03](https://www.youtube.com/watch?v=yQEE_bXt5DE&t=363s) Plaintiff Firm Adoption vs. Big Law Realities
Hamid contrasts the structured environments of corporate legal departments with plaintiff-side practice (e.g., personal injury, workers' compensation). Managing operations across 1,000 firms and 4,000 staff members reveals a lack of technical infrastructure that makes basic CRM adoption difficult, let alone advanced AI implementation.
In response to a follow-up question from John, Hamid confirms that this lack of internal oversight leaves plaintiff firms highly susceptible to AI hallucinations and rapid churn, as tools are adopted without proper checks and balances and quickly discarded when they fail.
> "There is no validation process for them. So they adopt them very quickly, they try it very quickly, and they drop it very quickly because there is no checks and balances in the system."
---
[07:52](https://www.youtube.com/watch?v=yQEE_bXt5DE&t=472s) Data Security, Governance, and Trust
The discussion moves to client confidentiality and attorney-client privilege. Emily outlines DragonGC’s architectural approach: adopting a zero-data-retention (ZDR) model where chat histories reside strictly on the user’s browser and documents pass through secure relays without being stored.
Blake reinforces that zero-data-retention is a mandatory baseline for regulated financial institutions. He shares an anecdote about a Wells Fargo attorney who wrote a 69-page prompt to analyze a complex legal document—a feat that earned him legendary status within the bank despite its high API token cost.
```
+-----------------------------------------------------------------+
| ENTERPRISE VENDOR CHECKLIST |
+-----------------------------------------------------------------+
| 1. Zero Data Retention (ZDR) | No server-side chat storage |
| 2. Enterprise SSO | Centralized access control |
| 3. SOC 2 Type II Certification| Verified third-party audits |
| 4. Data Encryption | In-transit and at-rest standards|
+-----------------------------------------------------------------+
```
- **Vendor Evaluation Criteria**: SOC 2 Type II compliance, enterprise SSO integration, and end-to-end data encryption.
- **Prompt Engineering at Scale**: The "69-page prompt" story highlights both the growing technical sophistication of in-house counsel and the need to monitor token costs (`/cost`).
---
[12:19](https://www.youtube.com/watch?v=yQEE_bXt5DE&t=739s) Automating Commoditized Tasks and Shifting Attorney Roles
Panelists examine how AI shifts daily legal work away from repetitive manual tasks. Eric breaks down Goodwin's tool portfolio allocation across commercial platforms, foundation model chats, and custom-built software.
Emily notes that automating routine document gathering enables junior associates to perform substantive legal analysis earlier in their careers. Blake adds that across legal, public affairs, and audit, the overarching objective is finding scale opportunities to remove unnecessary human labor.
- **Junior Associate Growth**: Shifting from manual comment-letter aggregation to higher-level analysis of SEC accounting disclosures.
- **Corporate Cultural Shift**: In-house legal teams focus on throughput and efficiency rather than billable hours, eliminating time-tracking overhead.
---
[16:15](https://www.youtube.com/watch?v=yQEE_bXt5DE&t=975s) Combating AI "Slop" and Quality Degradation
The panel addresses the risk of unverified AI outputs—termed "AI slop"—referencing recent high-profile incidents in management consulting where firms generated verbose, low-quality reports using AI. Emily warns against the temptation to pad work product or use AI as a substitute for critical thinking.
She shares a cartoon anecdote illustrating the absurdity of using AI to inflate five bullet points into a 12-page paper, only for the recipient to use AI to condense that same paper back into five bullet points.
> "Chat GPT take five bullet points, turn it into a 12-page paper. And then the next scene was somebody saying, I don't have time to read this 12-page paper, turn it into five bullet points."
- **Quality vs. Quantity**: Re-evaluating legacy billable-hour incentives that reward length over concise, actionable advice.
- **Analytical Rigor**: Ensuring that AI tools enhance efficiency without eroding the foundational thought process required for sound legal counsel.
<ElicitationsGroup message="Explore related topics from this panel:">
<Elicitation label="Summarize Part 2 of this transcript" query="Summarize Part 2 of the AI-Enabled Lawyer panel transcript."/>
<Elicitation label="Compare law firm vs. in-house legal AI strategies" query="Compare how law firms and in-house legal departments evaluate and deploy AI technologies based on this panel."/>
</ElicitationsGroup>
## Video description
Blake Rooney - AI Business Executive - Managing Director, Wells Fargo
Eric Tan - Chief Digital & Technology Officer, Goodwin
Emily Drazan Chapman - Legal AI Architect, DragonGC
Hamid Kohan - Founder / CEO, Practice AI
John Buyers - Partner, CMS LLP
This panel features in-house counsel,law firm leaders, and AI leaders using AI to reduce time spent on low-value work and increase precision in high-stakes matters. Attend this panel to understand how the cutting edge of AI powered legal work will shape the legal industry moving forward. Topics include contract review automation, AI-driven litigation analytics, and new ethical concerns around confidentiality and malpractice in AI-assisted lawyering.
AI 赋能律师：从调研检索到风险管理
当前法律机构正推动一线律师深度参与定制化 AI 工具与 Agent 的开发，但若缺乏成熟的技术基建与校验机制，AI 幻觉易导致工具被迅速弃用。与此同时，零数据保留等数据安全标准已成为严监管机构筛选供应商的硬性底线，业内专家亦提醒防范未经核实的“AI 废话”，并敦促反思奖励冗长内容的传统计费工时模式。
Goodwin 设立“Goodwin Builders”项目，组织兼具计算机背景的律师与工程师协作开发专属并购及风投工具；Wells Fargo 亦直接让一线律师深度参与定制自动化 Agent 的设计。 (3:00)
Practice AI 创始人 Hamid Kohan 指出，原告律所由于缺乏成熟的技术基础设施与审核校验流程，面临严重的 AI 幻觉风险，导致工具在快速引入后往往被迅速弃用。 (6:03)
数据安全与保密性成为法律 AI 落地的关键底线，零数据保留（ZDR）、SOC 2 Type II 认证与企业级 SSO 已成为金融等严监管机构筛选供应商的硬性标准。 (7:52)
专家对未经核实的“AI 废话”（AI slop）发出警告，敦促律所避免利用 AI 凑字数填充工作成果，并反思奖励冗长内容而非精炼实用建议的传统计费工时模式。 (16:15)
Reading summary
Legal AI adoption differs sharply across corporate legal departments, large law firms, and plaintiff practices, shaped by distinct technical infrastructure, billing models, and operational capabilities. Panelists emphasize that successful implementation hinges on zero-data-retention security, purposeful workflow redesign, and resisting the generation of low-value, verbose outputs.
Divergent Deployment Models Across Firm Types (3:00)
Adoption strategies reflect stark divides in operational readiness and organizational scale. Eric Tan reported that Goodwin allocates its legal technology portfolio across commercial vendors such as Harvey and Lawroid (50%), foundation models including Claude and Gemini (30%), and custom in-house software (20%) developed by 'Goodwin Builders,' a program pairing software engineers with computer science-trained lawyers. Blake Rooney noted that Wells Fargo similarly combines commercial platforms with natively built internal agents co-designed by line attorneys.
Infrastructure Gaps in Plaintiff-Side Practices (6:03)
In contrast to corporate legal departments, Hamid Kohan explained that plaintiff-side personal injury and workers' compensation practices often lack the basic technical infrastructure or CRM systems required for effective AI deployment. Based on operations across 1,000 firms, Kohan stated that the absence of structured validation frameworks leads these practices to adopt AI tools rapidly, only to abandon them when hallucinations occur without internal checks and balances.
Data Security and Enterprise Risk Controls (7:52)
Enterprise adoption requires strict privacy architectures to preserve attorney-client privilege. Emily Drazan Chapman outlined DragonGC's zero-data-retention (ZDR) model, which keeps chat logs exclusively in user browsers and processes documents through secure relays without persistent storage. Rooney confirmed that zero data retention, enterprise single sign-on (SSO), end-to-end data encryption, and SOC 2 Type II certifications are non-negotiable baselines for financial institutions, while noting the operational need to track API token expenses as internal attorneys develop complex prompts.
Task Automation and the Risk of 'AI Slop' (12:19)
Automating routine document collection allows junior associates to undertake complex SEC disclosure analyses earlier in their careers, shifting institutional emphasis toward throughput over manual billable hours. However, panelists cautioned against quality degradation, or 'AI slop.' Chapman warned against replacing analytical rigor with padded drafts, illustrating the inefficiency through an anecdote of an attorney expanding five bullet points into a 12-page memo only for the recipient to summarize it back into five points, challenging legacy billing incentives that reward document length.
While institutional enterprises and large law firms are building sophisticated custom agents and ZDR frameworks, structural divides leave smaller practices vulnerable to tool churn and output errors. The legal sector faces the unresolved challenge of reconciling AI-driven compression of work with business models that historically incentivized billable volume over conciseness.
[0:04] I'm a partner at CMS, an international law firm.
I'm the kind of intended demographic for AI
[0:09] enablement. Uh probably the worst kind, a partner
that uh doesn't really uh take any notice of
[0:15] uh AI uh roll out within their their own practice.
But there you go. Percher turned gamekeeper. And I
[0:20] advise typically clients on AI. And I've got
a very um esteemed panel of guests with me.
[0:26] and we're going to leap straight in because
I'm conscious of time and we've got a lot to
[0:30] get through. Uh so I guess the first question I'm
going to ask you all in order really is to start
[0:38] with a brief two sentence introduction from all
of you detailing the most critical AI priority
[0:44] on your lo your legal road map for this year.
So Hamid do you want to kick off? Sure. Thank
[0:50] you. [clears throat] My name is Hamid Kohhan from
the law practice.ai. We develop AI for a plenty
[0:56] firms which is very different than everybody else
here. Uh so we have a whole separate different
[1:01] challenge on implementing AI as a applies to
the business of law more than a practice of
[1:07] law because it's mainly operational AI from intake
to document collections to drafting and so forth.
[1:14] And the challenge is the implementation not AI
itself. Thank you. Eric, my name is Blake Rooney.
[1:22] I'm a managing director at Wells Fargo. I'd
spent the prior 20 years with a large law firm,
[1:27] Hush Blackwell. Um, the two biggest challenges for
us are just how do we drive efficiency and how do
[1:32] we better manage the relationship with our outside
law firms due to artificial intelligence. Hi,
[1:40] I'm Emily Chapman with Dragon GC. Uh, we are a
legal AI platform that focuses on SEC disclosure.
[1:47] And the biggest thing that I'm focused on this
year is addressing the gap between what AI can
[1:52] do in a legal context and what lawyers actually
trust. And so the big thing is that we're building
[1:59] workflows that are auditable and verifiable so
that attorneys can keep doing the work that they
[2:05] care about without moving blind. Eric, thank you
John. Um, first of all, I want to say thank you
[2:12] for being here instead of Black Hat. This is a
far better event than Black Hat. Um I'm Eric Tan
[2:19] um chief technology officer uh from Goodwin. We're
3,000 person uh law firm. Um and I'm 10 months old
[2:29] uh into the legal industry. Prior to this, I
was building drones for law enforcement uh and
[2:35] building enterprise software companies. Um the one
thing that I'm laser focused on is reimagining how
[2:41] um you know we can uh improve our client
experience. Uh if I think about it stepping
[2:47] into the legal industry I feel I've walked back
20 maybe 30 years at times and I just see a future
[2:54] where we could just walk up to a client use a QR
code and get you know legal work done. Thank you
[3:00] very much Eric. And well, now what we're going to
do is uh talk about specific AI initiatives. And
[3:06] I'm going to leave Hammed out for the moment.
We'll pivot back to him. Um we'll start with
[3:11] Blake. Really, can you walk us through a specific
AI initiative your team's recently launched or
[3:16] is currently building out, focusing on the
exact problem it's attempting to solve and
[3:21] the measurable impact it's having on your legal
operations? So for us, it's around contract and
[3:28] agreement analysis. um you know with the bank with
a bank with 200,000 employees there's quite a bit
[3:32] of work that goes through the legal department
um and it's interesting we've we've spent a lot
[3:37] of time looking at the big you know the Lora and
Harvey platforms and at the same time we've got
[3:42] capabilities that I'm unaccustomed to coming from
a law firm with 2,000 people coming to a a large
[3:49] organization with 34,000 people in information
technology we can do a lot in a hurry um and we've
[3:54] seen that accelerate over the the five months
that I've been with the organization So, um,
[3:59] we're starting to go through and and see what we
can do natively. Uh, we're building agents. We're,
[4:05] you know, using the attorneys expertise. We're
getting them involved in the construction of the
[4:10] um of the agents. It's been refreshing to kind of
see how fast we can move uh when we put our minds
[4:16] to it. That's fascinating. Thank you, Emily. So,
we're trying to address the mechanical things that
[4:23] go into lawyering. Uh if you are looking to figure
out who you should be telling a NOMGV committee
[4:30] to choose for your next director, historically
you had an associate going through your director
[4:36] skills, the director skills of other people in
your industry and figuring out where do we have a
[4:41] skills gap and that just is hours and hours of
research. And what our platform allows is the
[4:47] ability to use AI to look not only at what other
skills people have, but what are the underlying
[4:53] requirements for the exchanges and the SEC. So
taking a mechanical process allowing the user
[5:00] to get the information more readily so they can
make the important decisions. Thank you. And Eric,
[5:07] um I guess the one thing I'm really excited
about that we've recently launched is uh
[5:11] Goodwin Builders program. It's attorney building
for attorneys. Essentially what we did was we
[5:16] identified attorneys who are really really good
in you know with a background in technology. So
[5:21] imagine a world where you've got you know uh
an an attorney from Harvard who is also a CS
[5:28] major who's now building capabilities within
the firm. But this individual is also really
[5:33] good in venture financing or you know mergers
and acquisitions and really you know giving the
[5:38] associates an opportunity to you know partner
directly with our software engineers and build
[5:43] you know capabilities within the firm. Oh
that's fascinating and we're going to pick
[5:47] up on that topic I think hopefully later on in the
conversation because I think adoption is a key key
[5:52] part of any law firm strategy uh whether you're a
defendant or a plaintiff law firm. On which note,
[5:57] I'm going to pass it over to Hamid to uh
to explain from his perspective. Thank
[6:03] you. [clears throat] So in the plaintiff system
is very different because through the legal soft
[6:07] system we have like 4,000 legal staff working
about a thousand law firms and sort of very
[6:14] certainly we have to train all of these folks to
be AI enabled. So we call them AI enabled agents.
[6:21] uh an adaptability of that into the PI firms or
employment firms or workers comps those kind of
[6:27] a plaintiff side is extremely challenging because
they don't have a structure or infrastructure to
[6:32] actually implement anything. Uh I remember
in the past just implementing a basic CRM in
[6:38] these firms was like crazy and that was very basic
straightforward CRM. Now you're adding AI on top
[6:44] of that is extremely challenging. So we have a lot
of case studies about implementing AI as part of a
[6:50] you know intake process or case management process
and so forth. Um it's a lot more challenging that
[6:58] big law or corporate law and so forth because
they have discipline and infrastructure. Most
[7:03] plaintiff firms don't have that. So it makes
it a lot more challenging. I'm just going
[7:08] to throw you a quick left field. I mean, if
you don't mind, do you think that means that
[7:13] um plaintiff law firms are more susceptible to
just taking these platforms on trust and therefore
[7:18] more susceptible to AI hallucination? Is that
the 100%. Um there's no validation process for
[7:24] them. So they adopt them very quickly, they try it
very quickly and they drop it very quickly because
[7:32] there is no checks and balances in the system
and they don't have the resources, the human
[7:36] resources in house to be able to actually do the
full-on implementation, training, verification,
[7:42] validation, compliance. There's none of that
stuff. So it's a lot more challenging. Well,
[7:48] that that thank you. And that leads us kind of
neatly into the next topic of of trust really.
[7:52] And I'm going to focus on um Emily and Blake on
this on this topic. I guess one of the the biggest
[7:59] barriers to AI adoption isn't legal capability,
it's trust. So, and before legal departments and
[8:04] law firms can realize the benefits of AI, they
need confidence that confidential information,
[8:10] privileged communications, and client data remain
protected. And Emily, I know you spent a lot of
[8:16] time thinking about how AI can be deployed
responsibly. In fact, we had a conversation
[8:20] about that last night in this in the speaker
room um in legal environments and Blake, you've
[8:25] evaluated these technologies from an enterprise
technology and governance perspective. [sighs] So
[8:31] let me ask you both um as client confidentiality
and attorney privilege are absolute pillars in
[8:38] the legal profession. How are you designing
AI workflows and choosing vendors to ensure
[8:43] sensitive case data is completely protected and
kept out of external training models? And Emily,
[8:50] why don't you kick off? So uh the model that
we've uh decided to move forward with is a zero
[8:56] uh data retention model. And that means that the
chats are not stored on our system at all. We keep
[9:01] everything on the browser of the user. Uh we do
allow internal information to be uploaded by the
[9:08] user, but that is then processed through a secure
relay system. And so the biggest thing that I want
[9:13] to say for the people that are here today is if
a vendor is telling you they have a zero data
[9:18] retention policy, that's wonderful, but you need
to push further and make sure that the information
[9:25] is being protected. And so you want to make sure
that the user uh that the vendor is confirming
[9:31] that they have data encryption in place, policies
and procedures that you have access to. Uh for
[9:38] example, we allow our users to access through
enterprise SSO systems and then from there
[9:44] uh we are uh able to have the systems verified
without having a separate credentialing system
[9:50] which will make your IT people much more happy.
Um, additionally, SOCK 2 compliance is something
[9:56] that means that not only the vendor has policies
in place, but they've been reviewed by a third
[10:01] party. So, those are really important things. Um,
and one of the key things that we've found that
[10:07] uh differentiates off vendors for law firms. As
you can imagine, being part of a large financial
[10:15] financial institution that's heavily regulated,
the the levels of audit and control mechanisms
[10:20] that exist there are pretty extensive. And
it's no different for anything in the legal
[10:23] department. And um we've got a you know model
risk management organization that spends a lot
[10:29] of time just analyzing how we do work um how these
platforms work and how we can work with them. Zero
[10:35] data retention is a non you know not having
that just takes you off the list immediately.
[10:40] Um we're we're you know just fanatical about
that aspect of it. And honestly, we've found
[10:46] that our capabilities to to make things happen
internally and build our own agents is kind of,
[10:52] you know, I would I would not have anticipated
that when I arrived that it would be that much
[10:57] of leaning towards what we can do internally. Um,
and obviously that streamlines things. We know,
[11:03] you know, the underlying infrastructure when
we're building the applications ourselves. Um,
[11:08] and the building of agents has just gotten so
fast. We had um an attorney who a few weeks
[11:14] ago emailed me and there was a particular legal
document that he wanted to uh to work with and
[11:19] he'd written a 69page prompt on how to build
that uh that got some attention. [laughter]
[11:26] He's now a legend inside the bank. So um but he he
had done it right. He had you know he' taken all
[11:32] the steps necessary to make sure that it produced
a good output. But that was a kind of a sea change
[11:37] moment for the entire legal organization when one
of their own, one of the attorneys who had been
[11:41] an application developer for 10 years before he
came to the bank had done something like that. So
[11:47] um you know the ability to do those things
internally has just you know really helped us kind
[11:51] of get some of those concerns off our plates. Like
69 page prompt. Talk about token maxing. I know.
[12:00] Yes. Yes. I asked him to do the slashcost after
he ran that prompt once and it was not a pretty
[12:07] number. So, and presumably [snorts] he's leading
on a prompt library at Wells Fargo now, is he or
[12:13] Yeah. Yeah. No, that's absolutely fascinating.
So, kind of moving on now. Um, and I think I guess
[12:19] we'll start with Eric on this one and then we'll
move down the line. But, um, we look at the impact
[12:24] of commoditized tasks being done by AI. So we hear
a lot about tedious tasks being picked up like
[12:32] intake and initial contract analysis and massive
document reviews. Um can you share a story about
[12:39] how automating these processes has practically
shifted the day-to-day responsibilities of your
[12:44] attorneys and legal staff? Yeah. So so the way we
think about you know automation at Goodwin is you
[12:52] know it's not build versus buy but you know maybe
a combination is how we like to think about it. So
[12:57] it's when I say build and buy. You know, similar
to what Blake shared, you know, we've invested
[13:01] quite a bit of time looking at um you know tools
like Lora and Harvey. I think they've done a
[13:06] really good job uh providing capabilities to our
lawyers. Um and you know that probably takes care
[13:12] of 50% of the workflows that we've seen. And I
think you know for certain teams um we give them
[13:19] access to a lot of the large um or the you
know the big three providers like claude uh
[13:25] enterprise chat GPT and Gemini um and that takes
care of you know 30% of the you know use cases or
[13:32] specific use cases that our attorneys have and it
is that 20% where you know we're building you know
[13:38] tailor purpose-built uh solutions for our clients
but that's kind of how we've you know broken up
[13:44] um you know kind of the work uh workflow within
the firm. [clears throat] Thank you Emily. Yeah
[13:52] so as I mentioned earlier you have really tedious
tasks of trying to pull information out and what
[13:58] we have found is the ability to take it that
you don't have somebody that is pulling the data
[14:04] building comparison tables and moving uh forward
and allowing the analysis to happen earlier. Um,
[14:12] so associates are able to gather information much
faster than they were able to do historically and
[14:18] then they're able to really start making good
judgment calls and we've been talking to some of
[14:22] the partners that we work with and they're seeing
that their junior associates are stepping up and
[14:27] you're no longer sitting there having somebody
pull every single comment letter that references
[14:32] a particular accounting standard. Instead, you're
actually getting analysis from the associate of
[14:38] what those letters are saying. and it's not just a
table. So, they're able to step up their work. So,
[14:45] it's interesting. So, coming from the law firm
side, we had a a particular practice area that
[14:50] that was a volume practice. And five or six years
ago, you know, pre-open AAI announcement of 2022,
[14:58] um we were already headed down the road of how
can we automate this work? How can we automate
[15:02] the in the intake and the management of this
particular set of work? And then it was part
[15:08] of my interest in moving to Wells Fargo was
just everything's at scale, right? I mean,
[15:11] when it's a 200,000ish person organization, so
everything you do has a scale factor to it. Um,
[15:18] that's important. So that's that's my job
right now is across the legal, public affairs,
[15:23] and audit organizations to go find out go find
all the scale opportunities and find out how we
[15:28] can drive out human effort. And we're seeing some
really good success across all three areas. Blake,
[15:33] does that mean that you're getting rid of
the chillables? Those were the best billable
[15:37] hours. You were just sitting there doing
document review. [laughter] So, okay. So,
[15:41] the one the one I will confess the one thing that
I didn't anticipate when I got to to a large law
[15:48] firm is that they don't track time, right? I mean,
and I me I mentioned it out loud in a meeting and
[15:56] just got glares and and looks of hatred like
we I work here for a reason. I'm not, you know,
[16:01] I'm not bu putting I'm not filling a time sheet.
I'm not putting a time sheet together. So, uh,
[16:05] we're having we're we're finding other ways to go
about trying to anticipate and track what they're
[16:10] doing. So, yeah. Um, before I move to Hammed,
just one sort of another left field that kind
[16:15] of struck me as you guys were talking and um,
one thing that I found really interesting which
[16:20] is not so much impacting the legal industry but
is impacting apparel industry. So, the regulated
[16:26] professional services industry such as the account
large accountancy firms uh, such as PWC and EY. I
[16:33] I don't know whether you guys have seen these
stories of um wholesale reports that are being
[16:38] generated by consultants within these businesses
and they're just full of AI slop. And there was
[16:45] one case in particular where um I think it was uh
EY no maybe EY in Australia did a report and they
[16:53] had to refund the Australian government something
like half a million Australian dollars because it
[16:57] was AI slop effectively. Do you see and that's
a symptom of brains being engaged in neutral uh
[17:04] and I know Emily this is your bug bear [laughter]
but do you do you see that as a challenge within
[17:12] your within your spaces and how do you how do you
counteract that? Yes, it is a challenge for sure.
[17:18] Um, right now, you know, I'm hearing from a lot of
people, I just want to push a button and get the
[17:24] answer and not have the underlying thought process
um, and the analysis that needs to be there in
[17:30] my opinion and the majority of people at this
conference's opinion. Um, but I think that the
[17:39] uh, somebody had a great cartoon that they put
up yesterday at one of the sessions which was
[17:43] somebody saying, "Oh my gosh, my 12-page paper
is due tomorrow. chat GPT take five bullet
[17:49] points turn it into a 12-page paper and then the
next scene was somebody saying I don't have time
[17:54] to read this 12-page paper turn it into five
bullet points and [laughter] that struck me so
[18:02] hard because the thing is we really we we want to
make sure that there is true efficiency but we're
[18:09] not just padding and the thing is is in in the
legal profession we've historically worked on the
[18:16] billable power and when AI sits there and takes
the time to cut things down, you try to figure out
[18:22] how how can I make it seem better. And so if you
sit there and your report was usually 20 pages,
[18:29] but now you can make it 65, that shows everybody
you worked a lot harder, right? Um, so yeah,
[18:36] we really want to make sure that users are making
a product that is useful both for their clients
[18:43] and internally because it waste people people's
time when you're just feeding generated things
[18:48] that are going off on tangents and not relevant.
[snorts] People need to be called out for it too,
[18:54] right? I mean, that's the key piece of the whole
thing. And you know I'm not an attorney but I
[18:58] occasionally in a presentation I'll put the
ABA model rules the duty of supervision and
[19:03] everything just to remind people that they
are responsible for the output. Um you know
[19:08] it's it the magic machine is not responsible
and accountable for that output they are. So
[19:16] sure one of the biggest realization we made
through this whole year has been creation of a
[19:21] whole new position called the AI implement. So AI
implementation person is reminds me of whatever 40
[19:28] years ago when we started having IT people because
we started having servers and and mainframes came
[19:35] in and so forth. So we had the IT person taking
care of the IT now is the AI implement AI manager.
[19:42] So this whole new structure doesn't matter if
it's a plaintiff firm or defense firm or a big law
[19:47] the AI implementers are the most key part of the
process with us. We have a we have [clears throat]
[19:53] a practice that is a consumer protection firm uh
that signs up retains about 700 clients a month
[20:01] and they had like 49 bilingual full-time intakers
and we had to go in there and implement the AI
[20:08] into the intake process. So we took like 18 of
these intakers out and we replaced it with three
[20:13] different AI intake. It lasted like 60 90 days.
We reversed it back into bringing the integers
[20:20] back into the system because it was affecting
the bottom line and that wasn't the thing. So
[20:25] we went back and hired two AI implementers just
for the intake so they can implement it properly.
[20:34] It wasn't just like they said it's not like
a turnkey solution. You just put AI in there,
[20:39] take the people out and everything goes the
same. It doesn't. It needs to be programmed,
[20:44] implemented, structured, and managed just like
a IT person would do. So now we're back into
[20:50] bringing the AI back in there and we made it per
department. So each department has an AI implement
[20:56] for that department. So if it's a case management,
document collection, eiscocovery, intake,
[21:03] setments, they all have their own AI person. It's
almost like a focal point of contact to implement,
[21:09] manage and maintain. They say AI stuff changes
every day and who's going to implement all the
[21:15] changes that they do. Can't even keep up daily
with every time you wake up everyone that this AI
[21:19] tool has changed a lot. So that implement needs to
manage it, implement it and train the people who
[21:25] are using it. I mean, I think you hit on a huge
point and it would be particularly opposite for
[21:30] someone like Eric who is kind of sitting over a
huge firm like Goodwin which is AI fatigue within
[21:36] the user base because the industry is moving
and fluctuating so quickly that it's not just
[21:41] train and forget install train and forget.
It's it's training every day and every week
[21:48] everything just changes so rapidly. Yeah. Yeah.
Yeah. And that's fantastic. I mean thank you. So
[21:53] I'm conscious of time. So we're now going to kind
of focus a little bit more on litigation which has
[21:57] its own interesting um considerations and really
around using AI in the area of prediction and and
[22:04] strategy using analytics to define [gasps] um uh
or to predict case outcomes. Um Eric are you able
[22:11] to kind of give us an insight into whether on how
you are doing that at Goodwin? It's a really scary
[22:17] topic. Um how should I answer that? Let let me um
I wouldn't say we're in you know out of all the
[22:24] areas that we're probably that's an area where
we're in experimentation phase you know being
[22:29] able to predict what a you know judge's outcome
would be. I think we've seen some capabilities out
[22:37] there that's interesting but it's you know given
everything that you just said around uh you know
[22:42] trusting that AI it's it's a really scary face um
or area for us. I mean our superpower is really
[22:49] in transactions. So we've spent a lot of time in
the areas of you know mergers and acquisitions,
[22:55] debt financing. I think litigation is an area that
we're you know uh approaching with caution. Um but
[23:02] again I think that there's a lot of data that we
we see a lot of analytics that we can share with
[23:06] the the teams as they go into depositions but
we're cautiously optimistic on you know kind of
[23:13] uh doing predictions of uh you know litigation
cases. Are you driving do you have your own
[23:17] sandboxing in-house teams within within Goodwin
that are actually We do. So we've got you know
[23:22] we've got an internal platform called Regina named
name named after our chairmeritus. Um and it is a
[23:28] secured environment like a sandbox environment
where attorneys get to play with. Fantastic.
[23:34] So Blake Emily do you have anything to to add
to that? So going beyond just pure litigation,
[23:41] what's really great is uh you have the opportunity
now to look at how regulators think of things.
[23:46] So they might not be a judge, but the SEC will
issue comments on filings and things like that,
[23:52] FDA, patent office. You now have the opportunity
at your companies and for your clients to start
[23:59] analyzing those letters and figuring out how
do those letters impact what you're working on.
[24:05] And so you have an ability now to you don't
necessarily want to drink the ocean but you
[24:11] can find out what is the point of interest of your
regulators right now. So while not litigation uh
[24:18] you are able to spot the issues that are important
to the people that are overseeing your businesses.
[24:25] So we're looking at it from multiple perspectives.
Part of it is, you know, we've got a long history
[24:30] of uh in our our billing systems and things like
that where we've actually been able to collect
[24:34] data on cases and outcomes. Um, but we're looking
at it from two perspectives. How do you know what
[24:40] can we manage internally? What do we send outside
law firms? Who's got the right expertise for
[24:45] different matters and the and you know, works in
the right jurisdictions and things like that. Um,
[24:49] so it's just, you know, we're seeing a lot more
efficiency with it. we're understanding, you know,
[24:54] where we might in the past have spent $100,000 to
settle a $15,000 case, right? I mean, that those
[25:00] are the kind of things that big organizations
do and we're now we've now got the data and the
[25:05] technology and the artificial intelligence to
better understand all of that and try and drive
[25:09] those costs out. And did you have anything to
add to that? Yeah, absolutely. on the on our side
[25:16] uh when the senior attorneys using associates or
parillegals doing the drafting of the complaints
[25:22] and then they review it, edit it, add to it and
so forth. There's been a big shift from having AI
[25:28] actually do the draft. So they're not submitting
it without reviews, without edits and so forth,
[25:33] but they basically cut the associate or parillegal
work out of it by taking the entire discovery,
[25:40] uh, analyzing it, do a case summary, do the draft
of the complaint, but they have no right to submit
[25:45] anything. So it goes to the attorney for review
and edit and then they submit. So that has been a
[25:51] big shift and speed up the litigation process by
months. Interesting. Well, it's clearly clearly
[25:58] a very fluid time. And now we're going to move
into one of the mo more kind of interesting and
[26:02] intractable intractable parts of the discussion
which is the impact on the billable hour which
[26:08] is kind of like the uh [snorts] the horror that
private practice law firms have of implementing
[26:15] AI through through their infrastructure and how
the tension between how to preser how to introduce
[26:20] operational efficiencies and how to maintain
u premium pricing um for for their work. And
[26:27] I'm going to I'm going to put put uh this this
position to um Hammed and Blake for alternate
[26:35] um views. I don't know who wants to Hammed,
why don't you kick off from a from a plaintiff
[26:40] perspective and then Blake if you can pick it up.
Well, since we don't have a billable hours in the
[26:44] plaintiff side is essentially is all about cost
reduction, expediting the process and reducing
[26:51] the resources especially human resources basis is
contingency billing. contingency billing because
[26:57] essentially they they just want to get paid in
the end and the less time and less cost it is
[27:03] to get there the better off they are. So they're
not into uh as much of a compliance and so forth.
[27:09] They just want to speed up the process. So they
are turning the first turn with us was turning it
[27:15] into virtual staff virtual legal staff to reduce
the cost for the plaintiff size and then it was
[27:20] AI on the top of the virtual staff to be able to
even expedite it further. Doesn't matter if it's
[27:25] a pre-lead or lead is the same process for the
plaintiff firms. So for a large organization like
[27:33] Wells we're you know we're taking a a a hard look
at the outside law firms that we work with. It's
[27:38] a large panel today and we're having very direct
detailed conversations about how they're using
[27:44] artificial intelligence and how we can share in
the benefit. So um what I would anticipate is that
[27:49] panel will get dramatically smaller over the next
couple of years as we consolidate work on firms
[27:55] who you know can efficiently and effectively
do the work using artificial intelligence and
[28:00] are transparent with us on what they're doing and
how they're doing it. Um so that we've got a good
[28:04] understanding that things are being handled in a
responsible way. Um but we'll definitely see fewer
[28:10] outside firms within the next couple of years.
Yeah. I mean it's a very interesting dynamic,
[28:14] isn't it? And it's often presented as a as a kind
of zero sum game, which is that if you put AI into
[28:19] a into a large law firm, then automatically the
the you know, the charging rate must come down.
[28:25] Um but I'm, you know, I'd say this because I am a
private practice partner. So, but I think uh there
[28:31] is a way to maintain premium pricing if you are
selling to the client. human judgment and you're
[28:37] using AI to kind of operationalize other parts of
the world. Eric, I don't know whether you you have
[28:42] any thoughts on that. I do. So, um if you look at
Goodwin, we're a $3 billion law firm. 20% of um
[28:50] you know, our revenues are actually fixed fees or
alter alternative fee arrangements. And you know,
[28:56] I don't think the billable hour will ever
go away to be honest with you. Um I I think
[29:01] there'll be cases like an an antitrust case where
the billable owl makes sense. I think where there
[29:08] you know like you know sim with what Blake said I
think where there's commoditized or you know kind
[29:13] of repeatable task that's an opportunity there
where we we're looking at using AI to make that
[29:19] um you know faster. Um, I don't necessarily
think, again, having done, you know,
[29:25] 13 transactions myself on the client side, I
don't think I'm I'm necessarily looking for the
[29:30] cheapest provider because the value I see when
I'm doing a transaction is speed. So, in fact,
[29:35] I might even pay more, right, if you can get out
of my way within, you know, 14 days as opposed to
[29:41] 30 days. So, I think that that there's something
there. Now, it's and that's how we think about it.
[29:46] I don't think again the billable hour billable
hour will totally go away but I think that mix
[29:51] you know I think that composition would you know
most likely change in the next you know call it
[29:55] two to three years. I I think that's a that's a
fascinating point. I mean obviously we touched on
[30:00] fixed fee but are you investigating other billing
metrics within the and how to monetize the the IP
[30:08] and the assets within the firm. Um because that I
see as another challenge coming out of AI adoption
[30:16] within law firms. We we we are and a couple things
that we're thinking thinking hard about and having
[30:20] open discussions with our clients um are you know
percentage of the transactions similar to what the
[30:26] investment bankers do right I think um you know
similar to you know what what Blake said we we sit
[30:32] on a wealth of data where we've we we can predict
with a good level of accuracy you know kind of how
[30:38] many hours our teams are spending on a particular
transaction so that's one thing we're looking at
[30:43] we're explor exploring a model where you know
we're giving clients and um you know access
[30:48] to uh insights uh historical insights um into
our data and that's you know on a subscription
[30:55] fee basis. So we're experimenting with a lot of
different things that you know a traditional law
[30:59] firm I I would imagine would not. I also think
there there is going to be a AI billable hours
[31:05] as well because all of these companies investing
tremendous amount of resources acquiring them,
[31:12] training them, controlling them, managing them,
train there's a lot of investment going on into
[31:18] every large corporations into the AI especially
the law firms. Isn't that just amortizing the
[31:24] cost of the license fee back to clients though?
I mean and law firms because it's an ongoing
[31:28] cost and it's expanding cost. is not just charging
them for the servers or anything like that. You're
[31:34] constantly investing just like you were taking
associate and it start with the $400 billable
[31:39] hours and then you're training this associate
to be a senior attorney. Now you're charging
[31:43] $3,000 an hour for the same person is the same
person but you're recovering your investment of
[31:49] five years of training the associate. Now that
associate is replaced with AI you're doing the
[31:54] same thing investing money and resources and you
need to monetize that somewhere down the line.
[31:59] So I think in the future there's going to be AI
billable hour based on the different firms on
[32:04] their investment on the AI and trying to monetize
that. Yeah, I don't know if we'll call it an AI
[32:10] billable hour, but I agree with you. Something
that we're watching really closely and we've seen
[32:15] a trend creep creep up over the past I want to say
maybe 3 to four months is just the you know the
[32:21] the consumption of you know the uh LLMs, right?
So you're seeing we're seeing associates consume
[32:27] anywhere between you know a,000 to $2,000 per week
which you know if you you know we've got 1,800
[32:34] attorneys it it adds up right so it's something
that we're starting to think a lot and and monitor
[32:39] really closely right Emily I'm genuinely curious
can I just ask the audience raise your hand if
[32:47] you would pay a bill that had an AI fee line on
it. Okay. This is just the beginning. No, no,
[32:55] no. And obviously you're here at AI4, but I it's
it's I've seen a couple people go like this. Um,
[33:02] but no, that that's great. Thank you. I mean,
it's it's a it's a it's a it's a tough challenge
[33:07] because, you know, when you look at token cost and
compute, it's an expensive resource for a law firm
[33:12] to to use. Uh, and the clients simply can't assume
that fees are going to go down as a result. But,
[33:17] you know, it's it's not as simple as that.
We're kind of I'm conscious of time and um Eric,
[33:23] your point kind of neatly segueed us into the last
last uh section which is really adoption. You were
[33:28] talking about associates using AI and I've heard
stories of um some law firms that provide bonus
[33:35] incentives on the basis of the number of prompts
they they give to AI or the number of you know
[33:41] uses of AI whatever metric they deploy. uh which
sounds an interesting tactic, but what are you
[33:48] doing to drive adoption? And I'll come to Blake
and Emily and Jam as well, but what what are you
[33:53] doing to drive adoption? Because adoption is the
key to success successful AI strategy. It's not
[33:58] just giving them the tools, taking them to the
uh taking them to the river. It's getting them
[34:03] to drink as well, isn't it? Uh absolutely. So,
we've got an So, competition competition is the
[34:09] way we do it. Uh we've got an internal program
called AI for all where we've uh provided access
[34:16] to not just our 1,800 attorneys but all 3,000
employees and we transparently publish every month
[34:23] uh kind of you know where where we're seeing uh
AI being adopted and the goal for us over the next
[34:30] uh 12 months is to save a million hours
across the firm. So, it's become this healthy
[34:35] competition that we're seeing across different
teams in terms of creative ways to use AI to
[34:41] um you know to to improve their task both in
the practice of law and in the business of law.
[34:48] Emily, did you have anything to add or uh really
really just the ability to have the verifi the
[34:56] information verified um is the biggest thing
that we've been seeing driving adoption is as
[35:02] soon as you're able to source the information and
be able to click in to see exactly what the AI has
[35:07] looked at to make that conclusion that is getting
the chatter among our clients and we're getting
[35:13] more people asking to get seats. Um, so the the
underlying data that is being used in order to
[35:21] run these programs, I think is really important
from an adoption perspective, particularly with
[35:26] the riskaverse crowd that is attorneys. [snorts]
Blake, clearly you've got some very enthusiastic
[35:32] people at Wells Fargo. If they can put together
a was it a 67 page prompt or something? 69 pages.
[35:39] Um, yeah. Yeah, I mean I think part of it and
part of the reason that I took the role at Wells
[35:44] Fargo was um that it's very much driven from the
board and the chair and my boss Salivan Bearden
[35:52] who's in charge of all artificial intelligence
for the for the bank. Um but there there's you
[35:58] know no misconception about what the mission
is which was important to me. I mean, it was
[36:03] um you know, everybody knows every day and the and
you know, you walk in the halls, you you hear the
[36:09] talk all the time and that was important to me.
It's just um you know, you're not pushing the rock
[36:14] uphill. I mean, it's there's constant enthusiasm
about it and it's and anecdotes. There's a person
[36:19] that I could tell that was pretty skeptical about
artificial intelligence and pretty skeptical about
[36:24] why I was there. And um over the last four months
I've heard a couple of conversations that you know
[36:32] she's she's a convert now right I mean she's used
the tools she understands the capabilities and
[36:39] um she's now able to spend the time on the you
know the intellectual tasks and not the you know
[36:44] not the lightweight analytical and clerical tasks
that you know consume your day when you when you
[36:49] should be doing more important more value added
sorts of things. So, we're seeing we're seeing
[36:54] people just driven in that direction now. I
mean, people are enthusiastic about it. Um,
[36:59] you know, like non-disclosure agreements
we see by the thousands every year. And,
[37:03] um, the the attorneys have taken on this mission
of, you know, find putting together a prompt,
[37:09] the optimum prompt that meets the the obligations
for 95% of the non-disclosure agreement. So,
[37:14] we can quit wasting time on that. Um, and, you
know, it's it's organic coming from them. We're
[37:18] collecting these prompts, you know, using them
against each other, seeing where they might where
[37:22] the logic might contradict and working towards
this optimum solution. And it's it's coming from
[37:28] the lawyers. I I do find it fascinating speaking
as a um a private practice lawyer uh and we're
[37:35] driven in our you know in inner souls by the by
the billable hour. Um there I I I suspect amongst
[37:43] the legal community in private practice there
is a kind of inner guilt in using these tools
[37:48] because they're quicker and we're so used to kind
of burning up hours doing things which really we
[37:55] shouldn't be doing that I think there's this kind
of uh there's a tinge of guilt when we use them.
[38:02] I I can't describe it in any other way. You know
we should we we should we shouldn't be you know we
[38:06] shouldn't be guilty about it. But I think that's
something that drives private practice attorneys.
[38:11] Yeah. I mean, did you have anything to add before
we move into the next? Yeah, there's actually a
[38:14] whole new process going on right now where people
who using the AI to do something like Eve or
[38:20] Harvey or anything like that. Now there's a second
there just like building a the hierarchy of the
[38:26] reviews they're building AIs to actually evaluate
other AI outcomes just like a supervisor would
[38:33] do. So there's a three layer of evaluation. So one
draft comes out of one AI and the next AI kicks in
[38:40] try to find out what was wrong or why they didn't
do it right and the third one is like a supervisor
[38:45] after supervisor. So they even replacing the
supervisors in this case the level of the attorney
[38:51] and attorney experience because the LLM has been
trained to identify the issues that comes out of
[38:56] even up or Eve or Harvey and then corrected or
make recommendations just like the train. So now
[39:02] we have a hierarchy of AIS on top of each other.
Yeah. Interesting. Um so I'm going to move to
[39:11] uh Blake and Emily now and because I'm fascinated
from your kind of in-house perspective um you get
[39:17] to see a range presumably of outside council uh
and you get to judge qualitatively how they are
[39:25] implementing and using artificial intelligence
based tools. Um, and some of them no doubt are
[39:32] more efficient than others or more class leading
than others. What what kinds of criteria are you
[39:37] developing? And I know Blake, Wells Fargo is a
very sophisticated buyer of legal services. What
[39:42] kind of criteria are you evaluating your external
council by? [snorts] And um without naming names,
[39:49] have you have you had any warning signals
from any of your external council panel
[39:54] on their lack of adoption of AI or misguided
adoption? It Yeah, it it's interesting. I mean,
[40:01] there's been some kind of worrisome, you
know, conversations with ones that are,
[40:05] you know, they're more worried about the business
problem that this is creating than, you know,
[40:10] how to get to the endgame on this. So um we we
do see a little bit of that but a lot of them
[40:15] are aggressively you know over the last 18 to
24 months moving in the right direction. That's
[40:20] promising. Um you know large organizations will
differentiate the legal work right there's this
[40:26] 75 or 80% undifferentiated work that's the you
know these things that you know almost any law
[40:33] firm can do. You want to find the ones that can
do it the best and the most efficiently. Um but
[40:37] then the other 20% is the bet the farm litigation
or the bet the farm transaction that you want the
[40:43] best work you want the name you want you know
all those different things involved in it. So
[40:47] um I think you know large organizations will
become more selective about who they partner with
[40:53] uh who they give their dollars to and you
know have very open conversations with them.
[40:58] Um I was at Microsoft last June and it worked
talking with their legal department and they were
[41:05] talking about the fact that they wanted to um they
had a goal within 24 months to get away from any
[41:11] firm that wouldn't move to a to an alternative
fee arrangement. Um so in you know all that
[41:16] transparency included in it, they want to know
how the work's being done so they can trust them.
[41:20] Um so I think we'll see a lot more organizations
take that approach. No, that's fascinating. It's
[41:26] one to continue to watch. Emily, have you got
any insights? I mean, Dragon GC is a is a very
[41:32] sophisticated user of AI. I mean, to what extent
do you kind of benchmark your external council?
[41:39] So, um, our external council does use AI. Um, and
they are very big fans of it. Um, the what I will
[41:48] say is just more the larger legal market. Dragon
GC came onto the scene approximately three years
[41:54] ago and three years ago people were saying AI for
legal work no can't possibly be done and I just
[42:00] got back from a conference last month where it was
probably 500 attorneys and the entire conference
[42:07] was how are we putting AI in and what are we doing
and that was a mix of in-house and law firms. So
[42:13] just the mental shift that has occurred, I'd say
over the last 12 to 18 months has been a gamecher.
[42:21] And so if you had that conversation longer than 6
months ago, have it again. Thank you very much. So
[42:29] I'm really conscious of time now. We're going to
um round off the discussion and I'm going to ask
[42:34] you all to just give me a very quick prediction
on where you think um the future of AI generated
[42:40] legal content is going whether whether the
market is going to become more permissive
[42:45] uh and allow AI more autonomy in the generation
of it or whether the uh the market is going to
[42:50] become more defensive or possibly a hybrid model.
[snorts] um each from your own perspective and so
[42:56] they're asking you to just think on your feet
but just just a just a couple of words Hamid I
[43:01] think we're in the very early stages of this and
I think it's going to go very quickly like it did
[43:05] in the com era and the app era we gone through
multiple era but this one is just 10 times or
[43:10] 100 times faster than any of those things as far
as adoptability so I think we're just is a proof
[43:16] of concept it hasn't actually even been done this
is just a proof of concept to use AI within the
[43:21] legal and it's going to go expand much faster and
much deeper that we thinking right now. Thank you,
[43:28] Blake. I think from my perspective, I think
the you know the the platforms that everybody
[43:34] talks about every day, the Harveys and Loras
and and co-consils will have to uh they're
[43:38] going to have to work a lot harder to keep that
business and you know they're they're competing
[43:42] against the foundation models. We've looked
at you know Claude Co-works legal components
[43:47] and seen some interesting things there where we
can you know we can add to it and customize it.
[43:52] uh but it's it's to the benefit you know of the
law firms and of the large in-house teams this the
[43:58] competition among um these tool makers and and the
foundation models uh I see a hybrid moving forward
[44:06] uh Blake even during this conversation referenced
that you want the name behind the transaction
[44:11] so the fact that there are attorneys that have
worked at different government agencies or have
[44:17] done a really big deal in the past or they grew
up with you that is what they that that's what
[44:22] the businesses are going to care about and the AI
is just going to make those practitioners faster
[44:26] at what they're doing. And the last word to Eric,
um I I've AI fatigued. My my prediction over the
[44:34] next 12 months is we'll see a massive, you know,
consolidation of all these AI tools. I think the
[44:40] first quarter itself was like 120 new vendors.
And you know, if I put you put your yourself in
[44:47] my shoes, my focus is not evaluating 120 new
legal tech companies. Is trying to figure out
[44:53] how to make, you know, Blake a better um give
Blake a better client experience. Very well
[45:01] said indeed. Well, we're we're at time. I just
want to for you all to put your hands together
[45:06] and give our excellent panel a round a round of
applause. Thank you very much indeed. [applause]
