Law firms are moving beyond symbolic AI adoption toward the more difficult work of redesigning how legal services are delivered. Olivier Chaduteau, founder of a Paris-based AI-native consultancy, describes three stages: initial dismissal, widespread licence purchasing with limited use, and the current shift toward operational change.
The third stage is not mainly a technology decision. Firms must rewrite workflows, train lawyers, define standards, decide where human review belongs and reconsider a business model historically tied to billable hours.
AI can reduce time spent on document review, drafting and research. That efficiency creates value, but it also weakens the direct connection between hours worked and revenue earned.
Workflow redesign becomes a management decision
Buying a legal AI product is easier than deciding how it should change a practice. Each use case has different confidentiality, quality and review requirements. A first draft of routine correspondence is not equivalent to advice affecting litigation or a transaction.
Firms need practice-level operating models that specify approved data, tools and escalation. Lawyers should understand what the system can do, how to verify sources and when automation is inappropriate.
These decisions are political because they affect status, staffing and responsibility. Junior lawyers may perform less routine research, while partners must establish how expertise develops when traditional training work changes.
Chaduteau argues that lawyers are more likely to adopt AI when it removes low-satisfaction tasks and creates time for interesting work. Successful implementation should therefore measure quality and professional experience alongside speed.
Efficient delivery challenges hourly billing
If a task that previously took ten hours takes two, an hourly model captures less revenue even when the client receives equal or greater value. Firms can preserve existing billing and improve their cost-to-revenue ratio temporarily, or redesign services around value and outcomes.
Client pressure may force the transition. A new firm without legacy billing practices can offer faster service at a lower or more predictable price, pushing established competitors to respond.
Value pricing is not new to law, but AI makes the question more urgent. A firm needs evidence about what automation changes: turnaround time, consistency, error rates and the amount of senior attention available for consequential judgement.
Pricing should not imply that AI makes legal work effortless. Review, confidentiality controls, model evaluation and accountability still have costs. The opportunity is to align fees with client value rather than concealing efficiency inside time records.
Clients will demand proof of responsible AI
Corporate legal departments face pressure to demonstrate AI adoption alongside other business functions. They will increasingly apply the same scrutiny to outside counsel.
AI capability may become part of panel selection and pitch processes. Firms could be asked which tasks use AI, how client information is protected, where data is processed and what measurable effect the technology has on speed and quality.
Confidentiality and sovereignty are decisive. A law firm should know whether prompts train an external model, which jurisdictions store data and how access is audited. Human review must remain visible for advice carrying professional responsibility.
The firms that benefit will treat AI as a management programme before clients turn it into a procurement requirement. That means choosing operating models, investing in training and confronting billing incentives that may discourage genuine efficiency.
Legal AI is entering its operational stage because the tools are no longer the hardest question. The difficult work lies in changing institutions built around established roles and economics.
Technology can make drafting and research faster, but competitive advantage will come from how a firm converts that speed into better service. Practices that offer clear safeguards, evidence of quality and pricing clients understand will be better positioned than those that merely announce another licence.
