As legal AI continues its rapid evolution, this month’s Legal AI Monthly Round-Up featured George Hannah, Founder of Best Practice, alongside contributions from Julia Klingberg and Andy McDonnell, Vice President of Innovation and Legal Technology at Lex Mundi. The discussion examined the rise of agentic AI, evolving pricing models, the growing trend towards proprietary law firm platforms, and the governance challenges posed by unsanctioned AI usage across legal teams.

The Rise of Agentic Legal AI

Hannah opened the discussion by highlighting a major trend emerging across the legal technology market: the move away from traditional chat-based interfaces towards agentic systems capable of executing multi-step workflows.

Recent announcements from both Legora and Harvey demonstrate this shift. Legora unveiled its Agentic Operating System (AOS), while Harvey launched a suite of configurable AI agents designed to automate specific legal workflows. Hannah noted that the market is increasingly focused on systems that can reason, make decisions and execute tasks rather than simply respond to prompts.

The conversation also touched on Anthropic’s latest Claude developments, including enhanced reasoning capabilities and the introduction of varying “effort” levels, allowing users to control the depth of analysis generated by the model.

Claude’s Growing Legal Ambitions

The panel discussed Anthropic’s increasingly visible push into the legal market through Claude for Legal. Unlike specialist legal AI vendors, Claude positions itself as a flexible foundational platform that can integrate with existing systems including SharePoint, iManage, NetDocuments and contract lifecycle management tools.

Seamless integration was highlighted as an increasingly important differentiator. Hannah described Claude’s role as a potential connective layer that allows lawyers to bring together information from multiple systems within a single interface. Participant Julia Klingberg added that from an in-house perspective, the ability to connect AI tools with existing knowledge repositories and business systems is becoming increasingly valuable, particularly as legal teams seek to streamline workflows without introducing additional complexity.

Learning Through Building

Drawing on his experience as a solicitor apprentice, Hannah explained how experimenting with Harvey’s agent-building tools has accelerated his own professional development.

He described creating agentic workflows based on legal processes he had previously completed manually. The exercise revealed an important lesson: building effective AI workflows requires a deep understanding of the underlying legal process. Feedback from more senior colleagues exposed gaps in his assumptions and ultimately improved both the workflow and his own understanding.

Klingberg echoed the importance of practical experimentation, noting that legal professionals often gain the greatest value from AI when they actively test tools against real-world use cases rather than relying solely on demonstrations or theoretical capabilities.

The discussion reinforced a broader theme emerging across the profession: AI adoption works best when combined with strong legal expertise rather than used as a substitute for it.

The Pricing Debate

An audience question prompted discussion around legal AI pricing models. Hannah noted that both Harvey and Legora have publicly discussed moving away from traditional per-seat licensing towards consumption-based pricing.

While usage-based pricing may lower barriers to entry, participants highlighted potential challenges for law firms seeking predictable budgeting and cost control. Gannon also referenced examples of organisations actively monitoring adoption and removing licences from lawyers who fail to use the tools, underscoring the growing expectation that AI proficiency will become a core professional skill.

Klingberg observed that in-house legal teams face similar considerations, balancing the desire to encourage experimentation with the need to manage costs and demonstrate measurable value from AI investments.

Kirkland & Ellis’ $500 Million Bet

One of the most discussed developments was the report that Kirkland & Ellis has allocated up to $500 million towards developing proprietary legal AI capabilities.

Hannah suggested this investment could further widen the gap between elite firms and the wider market by enabling highly customised AI solutions tailored to specific legal workflows and client needs.

Andy drew comparisons with Allen & Overy’s early relationship with Harvey, arguing that Kirkland may be following a similar path by leveraging proprietary data and expertise to create a competitive advantage. The panel explored whether such platforms might ultimately be commercialised and offered directly to clients as part of a broader service offering.

The discussion highlighted a growing strategic question for firms: whether to build, buy, or partner when developing AI capabilities. Klingberg noted that many corporate legal departments are asking similar questions as they evaluate whether to rely on external vendors, develop internal capabilities, or adopt a hybrid approach.

Shadow AI and Governance Risks

The conversation concluded with findings from a recent Access Group survey suggesting that many lawyers and paralegals continue to use unapproved AI tools despite organisational policies.

Hannah suggested that limited access to approved enterprise tools may be driving some of this behaviour. Klingberg highlighted the importance of providing lawyers with approved, user-friendly alternatives, arguing that restrictive policies alone are unlikely to prevent unsanctioned usage if teams feel their needs are not being met.

The findings reinforce the importance of balancing governance, security and accessibility as organisations continue their AI adoption journeys.

Looking Ahead

The session closed with agreement that the legal AI landscape continues to evolve at extraordinary speed. As vendors race to develop agentic capabilities, law firms explore proprietary solutions and foundational models compete for market share, legal professionals face increasing pressure to understand both the technology and its practical application.

The key message was clear: experimentation, training and adaptability will be critical as the next phase of legal AI adoption unfolds.

AI Trends Shaping Legal Practice

• Legal AI is rapidly moving beyond chat interfaces towards agentic workflows capable of executing multi-step tasks.
• Harvey, Legora and Anthropic are all investing heavily in agent-based functionality.
• Pricing models across legal AI providers may shift from per-seat licensing to usage-based consumption.
• Major law firms are increasingly considering proprietary AI development as a strategic differentiator.
• Unapproved use of AI tools remains widespread within legal teams, creating governance challenges.
• Microsoft is quietly expanding its legal AI capabilities within the tools lawyers already use every day.