
Edition 17: Why most law firms saw little ROI from AI in 2025
Why legal AI didn’t pay off this year

Rok Popov Ledinski
Founder | MPL Legal Tech Advisors
Jan 1, 2026
MPL Legal Tech Advisors: The Legal AI Brief
Thursday, 1st January 2026 - 17th Edition
This Week's Theme
2025 was the first year many law firms could realistically evaluate AI based on experience rather than expectations.
The result, according to the ABA's Year 2 AI Task Force report, is clear:
Most firms experimented, very few measured impact, and even fewer saw sustained returns.
This issue looks at why.
Insight of the Week
AI was adopted faster than legal work was structured. The ABA report shows strong growth in AI awareness and trial usage, but weak evidence of operational payoff.
The common failure point was not model or technology quality, but rather structure.
Most firms introduced AI into environments where:
workflows were informal or undocumented
data lived across email, shared drives, and disconnected systems
responsibility for review and supervision was unclear
In those conditions, AI outputs could not be trusted, compared, or reused.Time savings remained anecdotal, while risk quietly increased.
The firms that reported benefits followed a different order:
They stabilized process and data first, and only then introduced AI selectively.
Legal AI in Action
🎬 The 5 Questions That Reveal If Your Firm Is Data-Blind
How firms make decisions without evidence and how to change that.
🎬 Turn Your Firm's Data Into A Decision Making Engine
4 week framework for moving from blind growth to profit.
Red Flag of the Week
The report highlights a shift that deserves attention.
AI is most often used in areas labelled "low risk": research summaries, drafting support, intake review, internal analysis.
In practice, these tasks are frequently chained together.
When early AI outputs feed later decisions without clear checkpoints, firms lose visibility into:
who reviewed what
which sources were relied on
where judgment was exercised
Looking at it from this lense, ROI was not the biggest problem in 2025.
Loss of control was.
What the Legal AI Frontlines Are Saying
In my talks with the leadership in law firms and in-house teams this week, the same explanations came up repeatedly:
1. "We expected the tool to organize the data for us."
2. "Usage varied so much we couldn’t assess impact."
3. "After early errors, confidence dropped."
The ABA report describes this pattern directly:
AI initiatives failed when treated as software deployments instead of operating model changes.
2025 marked the year when the technology moved faster than governance.
Looking Ahead
🎙 This Saturday at 2pm CET!
This week’s guest on Rok’s Legal AI Conversations is Vaishali Gopal, former lawyer, and now vice president and head of innovation at Counselect. Vaishali shares a practical approach to legal ops transformation: business alignment, process-first design, data/knowledge foundations, automation vs AI, contract playbooks, change management, and build vs buy for legal tech.
![]() |
Each edition of Legal AI Brief brings practical lessons from firms using AI safely.

Rok Popov Ledinski
Founder | MPL Legal Tech Advisors
Share





