Companies are under real pressure to do more with less, especially for legal and deal work. AI tools can help, but it changes the risk calculation and the tools do not always offer ‘savings’ in the way expected. And too often, the AI ‘savings’ conversation is just framed too simply.
AI can create real efficiencies. But “efficiency” is not the same thing as “automatic savings”. This distinction matters, especially in legal and deal work where speed only creates value if the output can be trusted. The ‘savings’ conversation is not simply whether an AI tool was used. It is whether the tool actually reduced the path to reliable work or just shifted cost and risk to human analysis.
The market tends to talk about AI in terms of inputs: Model costs, token prices, speed, subscription fees, and usage rates. But clients do not buy tokens. They buy analysis, risk management, outcomes, and completed work.
A cheaper AI tool is not necessarily cheaper if it produces work that requires significant cleanup. A faster first draft is not necessarily efficient if the human reviewer has to spend the saved time finding errors, checking assumptions, fixing tone, reworking analysis, or deciding that the output cannot be used at all.
A recent Pitchbook analysis illustrates this contextually by comparing token costs to the actual costs to complete the work: Some AI tools have better value, but not the best token price. Sometimes AI meaningfully compresses the work. Sometimes it improves quality. Sometimes it provides a strong starting point. And sometimes it simply moves the burden from creation to analysis. That burden is not incidental; it is where professional analysis lives.
So when clients inquire about AI use translating into reduced fees, the answer typically offered is ‘sometimes, yes, but not always.’ But, for me, this answer is incomplete. Pitchbook’s analysis aligns with my anecdotal recognition that, sometimes, an AI tool does not prove useful for a project (or minimally so) – whether that is from a cost-savings standpoint or added-value standpoint. And other times, the AI tool works well in assisting the overall workflow so that the analysis can be completed more quickly or more in-depth.
And that is where the AI ‘savings’ conversation should start: What value is the client obtaining through the attorney's use of AI tools? The answer to this question is the real advantage for clients.
In many cases, the value of AI is not that the same work becomes cheaper. It is that the same investment produces better work, faster issue spotting, broader coverage, more consistency, and more informed analysis. The question is not simply “Was AI used?”
If the AI output reduces the path to that result, there may be true savings. If it improves the result without reducing the required human analysis, the client may be receiving better quality for the same fee. And if the output is weak, the apparent savings can disappear quickly.
AI changes the economics of legal and professional work, but it does not eliminate the cost of analysis. It makes that analysis more important.
For in-house counsel and deal teams, the more useful conversation is not whether AI was used, but how it changed the path to a reliable outcome. Did it reduce cycle time? Improve issue spotting? Create better consistency across the workstream? Free the team to focus on higher-value judgment calls? Was there value derived from use of AI?
When it comes to the impact of AI on the work being done, this is where the conversation should begin.
If your team is thinking through how to measure AI’s impact on fees, workflows, or deal execution, I would welcome the conversation. The most productive discussion is not just where AI is being used, but where it is creating value that can be trusted.
