Our Take
Harvey has moved from productivity toy to operational backbone in legal services, but the claim that in-house teams see cost savings remains almost entirely unproven—only one firm reported actual savings despite 48% of them mandating AI use.
Why it matters
Law firms are betting hiring and partnership strategy on AI fluency, yet retention of headcount suggests efficiency gains are being swallowed by volume expansion, not cost reduction. In-house legal teams are getting squeezed in the middle: they mandate AI but see no savings to show for it.
Do this week
In-house counsel: audit your law firm contracts to specify what cost-sharing mechanism applies to Harvey-driven productivity, before the bill reflects the efficiency but your budget does not.
68% of law firms now run Harvey agents in production
RSGI, an independent research firm, surveyed 87 respondents across 60 law firms and 27 in-house legal teams between April and June 2026 (per the commissioned but independently-conducted study). The findings show steep adoption growth in just six months since RSGI's first Harvey report in November 2025.
Sixty-eight percent of law firms and in-house teams are already deploying Harvey-based AI agents. One-fifth (21%) of law firms are running more than 50 agents in production. Integration with document management systems, Microsoft 365, and MCP/API are cited as key enablers.
Power users—the subset of lawyers heavily relying on Harvey—save an average of 11 hours per week, up from 8.5 hours six months earlier (per the survey respondents' reports). Fifty-nine percent of firms able to track changes report increased lawyer utilisation; 44% report increased revenue; 53% report increased profitability among lawyers using Harvey.
The narrative has shifted. Fifty-five percent of law firms and 48% of in-house legal teams now describe Harvey as foundational infrastructure rather than a productivity tool. Fifty percent of organisations now have 30% or more power users (up from 34% six months ago). Fifty-seven percent of law firms and 70% of in-house teams say a lawyer can reach power-user status in under three months.
On hiring, 82% of law firms and 94% of in-house teams are changing recruitment criteria to prioritise AI skills. Twenty-three percent of law firms have embedded AI competency into formal partner appraisals; 33% for associates. Yet neither law firms (69%) nor in-house teams (68%) plan to reduce headcount over the next two years.
The cost-savings promise remains unproven where it matters most
This is where the story fractures. Sixty-one percent of law firms say they have proactively discussed Harvey efficiencies with clients. Only 4% of in-house teams report that their law firms have approached them about cost savings. More telling: 48% of in-house teams have mandated their law firms use AI, yet only one firm reports receiving cost savings as a result.
That single reported savings outcome, against 48 mandates, suggests one of two things. Either the efficiency gains are still too fresh to measure reliably, or they are real but being captured by law firms rather than passed to clients. Flat headcount plans at both law firms and in-house teams point toward the Jevons Paradox: increased efficiency is enabling more work to be done rather than fewer lawyers being hired.
The market dynamics are asymmetric. Law firms control the product integration, set the client communication, and measure the time savings. In-house teams bear the mandate but have no visibility into whether they are actually paying less. This creates a one-way flow of efficiency gain—captured by the supplier, not the buyer.
Nail down the financial terms before mandating the technology
If you are an in-house legal team and you have mandated Harvey use at your law firms, you need to audit what happens to the efficiency gains. The RSGI data shows that 47 out of 48 teams who mandated AI have not received documented cost savings. That is not a lagging metric—that is a missing contract clause.
Law firms are hiring for AI skills and treating Harvey as foundational. They are not shrinking headcount. If they are saving 11 hours per power user per week and using that time on more client work, the question is: who is billing for that extra work? If the answer is "the firm," then you are funding a productivity gain that your law firm captures, not one you share.
The speed of adoption is real. The time savings are real. The strategic pivot by law firms to AI-native hiring is real. The cost benefit to in-house teams remains almost entirely undocumented.