Agents, AI, and Ari..! – Artificial Lawyer

And here’s the last piece by Draftwise covering ILTACon on Day Three, which looks at agents and also AI Strategy. Plus, they held a fun run along with the energetic Ari Kaplan. (Maybe AL will be there IRL next year….)
 
–
 
Orchestrating Intelligence: AI Agents in the Legal Space – Session Summary

Speakers: Lisa Erickson (Aderant), Matt Zerweck (Harvey), Adam Ryan (Litera), Joel Hron (Thomson Reuters)

What Are AI Agents?

AI agents are goal-oriented systems that understand context, plan actions, and execute tasks autonomously. Unlike traditional AI that performs discrete functions, agents work like “a good co-worker” – understanding tasks and available tools, planning execution, and checking in for guidance when needed.

Key difference: You tell agents what to achieve, not what specific steps to take.

Why They Matter

Strategic Impact: Matt Zerweck noted agents “enable people just to do much more than they were ever able to do before… at a higher quality.” Joel Hron emphasized they “amplify the most human parts of the job and certainly the most difficult parts.”

Human Oversight: As agents become more autonomous, verification becomes critical. Future software will optimize verification speed rather than human-driven workflows, requiring transparent citation and source tracking.

Current Use Cases & Results

Email Processing: Agents proactively understand email context and execute actions like responding to experience inquiries or generating pitch materials.
Document Drafting: Customers report 50-70% time savings reaching early drafts with better consistency by embedding firm and client preferences.
Legal Research: “Deep research… is like the most profound example of agents” (Joel Hron), showing 60%+ time savings while discovering new arguments in cross-jurisdictional litigation.
Contract Analysis: Identifying standard terms, flagging non-standard provisions, and proactive risk identification across contract portfolios.

Training Agents: Three-Pillar Approach

Planning & reasoning capability – Core logical processes
Purpose-built tools – APIs designed specifically for agent use
Context provisioning – Access to relevant data (both proprietary and third-party)

Future Outlook (1-10 Years)

Ecosystem Development: Joel Hron predicts “ecosystems of agents that develop and communicate and collaborate with each other more effectively.”
Proactive Intelligence: Matt Zerweck envisions agents that “reach out to you” before being asked, suggesting actions based on incoming information.
Data Requirements: Adam Ryan emphasized that successful firms will have “really good structured data sets of their firm’s experience.”

Implementation Considerations

Give agents good information – The more context agents have, the better they perform
Start with easy tasks – Begin with simple, clear tasks before trying complex workflows
Check their work – People still need to review what agents do
Control access properly – Make sure agents only see data they should see
Results depend on the task – Some tasks need no help, others need frequent guidance

Key Takeaway

Joel Hron: “However big you think this is going to be in five years, it will be even bigger than that probably.”

The panel agreed that while technical capabilities exist today, the strategic transformation will unfold gradually as firms build proper data foundations and verification processes.
 
—

Session Two: Actionable AI Strategy & Policy
 
Speakers:
Sean Monahan – Senior Director, Data Modernization, Harbor (Moderator)
Sukesh Kamra – Chief Knowledge & Innovation Officer, Torys LLP
Christian Lang – Founder & CTO, Lega
Anna Corbett – Practice Innovation Strategist, Akin Gump

Audience Profile
Polling revealed 44 of 89 attendees (nearly 50%) are currently in the “piloting tools” stage of their AI journey, representing firms actively testing but not yet at full deployment

Key Debates and Findings
1. Strategy vs. Experimentation
Winner: Balanced Approach (Anna Corbett’s position)
Sukesh’s position: Start with clear strategy first – “You need a map. You need an objective, a goal.”
Christian’s position: Let strategy emerge through experimentation – “We have absolutely no idea where this is going.”
Anna’s winning argument: Balance governance with flexibility – “You have to be really prepared to be iterative and have a strong foundational governance policy that is sort of able to be flexible.”
2. Policy Development
Winner: Policy First (Sukesh Kamra’s position)
Sukesh: “We live in a regulated industry with unlimited liability as lawyers. We need a policy at the outset.”
Christian: Focus on structural safety over written policies – “Policies are only as good as they are operationalized.”
Anna: Let policies mature over time based on actual use.
3. Technology Investment Strategy
Winner: Strategic Platform Investment (Anna Corbett’s position)
Sukesh: Conduct readiness assessment before major investments
Christian: Avoid big investments, focus on R&D and experimentation layer
Anna: “Investing in those foundational enterprise AI tools that are going to immediately enhance daily productivity”
4. Transformational vs. Incremental Change
Result: Tie/Mixed Views

Christian: “We are easily within 18 months of technical legal skills not mattering for all intents and purposes.”
Anna: Plan for both short-term efficiency gains and long-term transformation
Sukesh: “It depends on leadership and culture within your organization.”

Lightning Round Insights

Technology vs. Politics
Consensus: Politics harder than technology
One dissenter noted: “The fundamental way that we all interact with AI are familiar with it is a barrier to adoption”

Chief AI Officer by 2026
Majority opposed – Sukesh: “AI is not something new… We don’t need to appoint a chief document management system officer.”

AI Ownership Structure
Divided views on whether one department should own AI

Anna: “I can’t imagine how it would succeed if it only existed on one team”
Christian: Need “point of strategic coalescence” but cross-functional approach

Key Takeaways
Three-Step Implementation Framework:

Conduct Readiness Assessment –
Evaluate change tolerance and structural readiness
Define success metrics (even if just usage tracking)
Assess architecture, risk management, and training capabilities
Deploy Foundational AI Tools –
Start with productivity-focused platforms
Help lawyers understand AI basics before moving up value chain
Balance experimentation with platform strategy
Implement Flexible Governance –
Establish baseline safety requirements
Avoid letting risk concerns block opportunities
Create structural safety rather than relying solely on policy compliance

Notable Quotes:
On adoption challenges: “The fundamental barrier to adoption [is] getting lawyers to use prompts” – Audience member
On change drivers: “Who gets it and who uses this are going to be the people who are going to drive the change more so than any one position.” – Christian Lang
On transformation: “Anyone who truly believes that this is incremental improvement technology and we plan on doing business for the next five, ten years the same way fundamentally that we do today, I think you’re going to be out of a job.” – Christian Lang

Conclusion
The panel and audience aligned on the nuanced nature of AI implementation, rejecting hard-line approaches in favor of balanced strategies that combine strategic planning with practical experimentation, all within flexible governance frameworks.
 
 —
 
And as mentioned, there was also a fun run, organised by Draftwise along with legal tech expert and consultant, Ari Kaplan, (in the orange vest above).
 
–
 
That’s all folks. Thanks again to the Draftwise team for their help covering ILTACon!

—

Legal Innovators New York, November 19 + 20.