If 69% of legal professionals are already using generative AI at work, but only 34% of firms have adopted legal-specific AI tools, the legal industry’s AI problem is no longer awareness. It is the gap between individual use and firm capability.
Those figures come from the 8am 2026 Legal Industry Report, and the governance numbers make the gap clearer: 54% of firms provide no AI training. Lawyers are learning through individual experimentation while many firms are still deciding what formal adoption should look like.
That creates a problem a new AI licence will not solve.
A law firm AI policy is necessary, but it is not adoption
A useful law firm AI policy establishes boundaries around confidentiality, acceptable tools, human review and professional responsibility. But a policy does not tell lawyers how AI should fit into a due-diligence review, research workflow or first draft. It does not train people, redesign processes or determine whether saved time produces any commercial value.
Ontario firms have another reason to take that distinction seriously. Since January 2026, Legal Aid Ontario requires roster lawyers to confirm annually that they have read and are complying with the Law Society of Ontario’s AI guidance. The professional obligations are not new; the expectation that lawyers understand how those obligations apply to AI is becoming harder to treat casually.
Firm-wide adoption therefore needs more than rules. It needs an operating model.
That means deciding which tools are approved, which workflows are suitable, who reviews output, what training people receive, who owns implementation and how results will be measured.
Saved hours are not a business case
The measurement question is particularly important.
Litify’s State of AI in Legal Report found that AI adoption reached 78% among respondents in 2025, yet fewer than 15% said AI was producing business impact.
That gap makes sense. Saving a lawyer two hours does not automatically create two hours of revenue.
As BLS explored in [Meta Just Made the Billable Hour Someone Else’s Problem — INSERT INTERNAL ARTICLE LINK], AI is forcing firms to confront a structural problem: traditional pricing and compensation systems reward hours, while AI is designed to remove them.
A successful AI workflow therefore needs an answer to a second question: what happens to the time you saved?
If it improves realization, increases capacity, accelerates turnaround or allows lawyers to spend more time on higher-value work, there may be measurable value. If nobody measures what happens next, “hours saved” is just an interesting statistic.
The next AI decision is organizational
Before buying another tool, firms should map where lawyers already use AI, identify the workflows worth standardizing, establish governance and training, and decide how success will be measured.
That is the real transition from experimentation to AI adoption in law firms.
BLS Consulting works with professional-services firms on AI strategy, technology, systems and process improvement at exactly this point: when individual experimentation needs to become a governed operating capability.
A firm-wide AI adoption diagnostic should start with one question: if your best AI user left tomorrow, what capability would remain inside the firm?
