This is a typical scenario. It shows how Remolda would approach AI-assisted contract review in a mid-size Canadian law firm, what we would build and what we would measure.
The situation
Picture a firm of a few dozen lawyers across several offices. Larger firms invest in technology and turn routine work around faster. Alternative providers compete on price for standard tasks. The firm competes on relationships and quality.
Associates spend a real share of their week on mechanical review: checking clause language against precedent, comparing contract versions, pulling key terms into client summaries and checking that statutory references are current. Clients question paying hourly for that work. Legal technology built for large firms needs innovation teams and budgets this firm does not have.
Many partners are sceptical of AI. A tool that seems to replace legal judgment will not be used, however good it is.
The approach
Audit (2 weeks). We sit with each practice group, follow live matters and map every document-handling step. Time logs give the baseline: hours per week on mechanical review by task type, and turnaround time on standard documents. Practice groups differ. Commercial real estate sees high volumes of similar leases; corporate work sees large diligence packages under deadline; regulatory work needs statute references checked.
The plan starts with the highest-volume group and expands only after lawyers trust the output.
Implement, wave 1 (2 months): commercial real estate. The system is configured on the firm's own precedents and clause standards for leases, purchase agreements, loan documents and title materials. For each document it:
- Extracts key terms such as parties, dates, amounts, conditions, renewal options and termination provisions into a summary template.
- Flags clauses that differ from the firm's clause library.
- Summarizes changes between versions when redlines are provided.
- Produces a review brief that the lawyer uses as a starting point.
Lawyers review the AI output and make every legal judgment. That line is explained to partners from the first meeting.
Implement, wave 2 (2 months): corporate and regulatory. Diligence extraction from corporate records, minute books, filings and financial documents, and a check that cited statutes and regulations are current, flagging references that were amended or repealed.
Empower (parallel). Sessions with lawyers and legal assistants as each wave goes live: what the system does well, where it misses nuance and how to check its output quickly.
What we would measure
Each item below is a target we would set with the firm and measure against the baseline from the audit. The size of each target is set only after the baseline is measured.
- Associate hours on mechanical review per week, by task type.
- Turnaround time on standard documents such as leases.
- Clause flags confirmed by lawyers, and non-standard clauses the system missed, from a weekly sample.
- Adoption: share of eligible matters in the practice group where the review brief is used.
- Fixed-fee readiness: whether review time is predictable enough to price standard reviews at a fixed fee.
Rules that frame this scenario
Lawyers stay responsible for the work. The Law Society of Ontario's guidance on generative AI and the Law Society of British Columbia's guidance both point to competence, confidentiality and supervision duties when lawyers use these tools. Client files never go into consumer AI accounts; the firm uses enterprise or API plans whose terms exclude training on client data, and records where each vendor processes it. Some courts, including the Federal Court, ask parties to declare AI-generated content in filings. The law firms page lists sources.
Key lessons
1. The lawyer keeps the judgment. The system does the mechanical pass and the lawyer decides. A partner who sees that division has a reason to support the system; a partner who feels replaced has a reason to resist it.
2. The firm's own precedents drive accuracy. Configuration on the firm's clause library and templates is what makes flags relevant. Accuracy is measured on the firm's own documents before go-live.
3. Predictable review time helps clients. Faster turnaround and a fixed fee for standard reviews are easier to offer when review time is measured and stable.
For related work, see AI for law firms, document processing and department AI training.