Canadian law firms have moved past asking whether AI belongs in practice. The question now is which tools meet the profession's standard for accuracy and accountability. CoCounsel Legal Canada, built by Thomson Reuters, pairs advanced AI with authoritative Canadian content from Westlaw Advantage Canada and Practical Law Canada. Thomson Reuters introduced a fiduciary-grade AI standard in May 2026, grounded in four concrete pillars: curated authoritative content, privacy and security by design, credentialled experts doing the work themselves, and outputs traceable to a verifiable source.
Fiduciary-grade AI is a formal standard Thomson Reuters introduced in May 2026 for AI used in high-stakes professional contexts. Thomas Sander, vice president of product management at Thomson Reuters, describes it as "a higher benchmark for AI used in professional contexts where accuracy, accountability, and trust are critical." The standard rests on four pillars: authoritative curated content, privacy and security built into the architecture, credentialled experts who do the work themselves, and outputs that trace back to a verifiable source. "It defines how AI should perform when professional liability is on the line," Sander says. For Canadian lawyers working under duties of care and regulatory oversight, that traceability is the difference between a plausible answer and a defensible one. No other widely available AI standard addresses all four requirements together in a single framework tailored to professional accountability.
The time savings are measurable and documented. Thomson Reuters' Future of Professionals 2025 report estimates AI adoption will free up close to 240 hours a year for the average legal professional. Sander says real-world results go even further: "Our Canadian customer data shows users are achieving more than double their speed on document review and contract drafting with CoCounsel Legal." The platform focuses on three high-friction task categories: complex legal research through Deep Research, large-scale contract analysis through tabular document review, and first-draft creation inside Microsoft Word. All three draw on Canadian primary law and practical guidance from Westlaw Advantage Canada and Practical Law Canada in a single pass, rather than requiring a lawyer to move between multiple systems and reconcile conflicting sources.
Generic AI tools were not built to distinguish Canadian law from American law, and that gap creates real professional risk. CoCounsel Legal Canada reasons specifically from Canadian content, so every answer reflects Canadian legal context rather than a blend of jurisdictions. Sander frames the core problem plainly: a capable model placed on top of unreliable sources can produce answers that are incorrect but sound so fluent that errors become difficult to detect. The content beneath the model comes from Westlaw Advantage Canada and Practical Law Canada, kept current by more than 1,200 attorney editors. KeyCite tracks whether a case is still good law. That editorial infrastructure, Sander argues, "is what makes the results defensible rather than merely plausible." That distinction matters when a client, regulator, or opposing counsel asks a lawyer to stand behind their work.
Tabular analysis works through a large volume of contracts at once, surfaces risk across multiple issues simultaneously, and produces draft reports at a scale that previously required a serious commitment of people and time. Drafting happens inside Microsoft Word, where CoCounsel draws on Practical Law Canada content and a firm's own precedents to build first drafts. In a common commercial litigation scenario, Sander explains, CoCounsel Legal runs a full contract set against several issues at once, with every finding linked back to the source document. That reviewable trail allows a lawyer, regulator, court, or auditor to assess whether the output holds up. "CoCounsel Legal makes the process extremely efficient, and each step in its planning, reasoning, and execution is visible to the user," Sander says. That visibility is what separates it from black-box automation.
No. Fiduciary-grade AI is built to support professional judgment, not substitute for it. Sander is direct on this point: "The goal is to compress the time between intake of an issue and a lawyer being able to make that final judgment." CoCounsel Legal surfaces research, flags risk, and links every finding to its source. The lawyer still chooses which issues to run, evaluates the results, and makes the recommendation to the client. That separation is not optional under Canadian professional responsibility rules, where a lawyer cannot delegate ultimate accountability to a tool or a third party. The platform makes that accountability easier to uphold by giving lawyers a transparent, reviewable record of how each output was produced, so they can explain their reasoning at any stage of a file.
The expectations are high and the gap between expectation and delivery is wide. Thomson Reuters' Future of Professionals 2026 report found that 96 percent of legal professionals say their AI must safeguard confidential data. Another 94 percent require verified authoritative content, and 90 percent need outputs they can explain and defend. Yet 41 percent of respondents lack access to professional-grade tools that actually meet those three requirements. Sander frames that gap as the core reason fiduciary-grade AI is a meaningful category at all: "Before professionals operating in high-precision fields can fully embrace deeper AI integration into their everyday workflows, they need to know that the AI they are using stands up to scrutiny and that its outputs are reliable and verifiable." The 2026 data makes clear that most current tools are not meeting that bar.
Thomson Reuters, headquartered in Toronto, pairs CoCounsel Legal Canada with a customer success team that works directly with firms through onboarding and beyond. Sander recommends that skeptics start with a recently concluded matter where the outcome is already known, then run it through CoCounsel Legal and compare the result against the original work. "Seeing how CoCounsel Legal does work that historically took days of manual effort in mere minutes wins over even the biggest skeptics," Sander says. Thomson Reuters is also building toward agentic AI: systems that respond to a plain-language question by forming a theory, executing a plan, and delivering finished work product. That direction will always be grounded in authoritative Canadian legal content, keeping the fiduciary-grade standard intact as the capabilities of the platform grow.
Thomas Sander: vice president of product management, Thomson Reuters; leads product strategy for CoCounsel Legal Canada and the Thomson Reuters professional AI platform.
