Jun 22, 2026 |

Building Fiduciary-Grade AI™ for Wherever Work Gets Done

Joel Hron  Chief Technology Officer,

Professional work has never happened in a single system. It spans research platforms, drafting tools, document systems, enterprise workflows, customer data, and now a growing set of AI-powered assistantsand agentic systems.

For ourcustomers,flexibility is powerful. Work can begin intheenvironment that is most natural for the task at hand. But it also creates a critical challenge: as work moves across more tools, models, partners, and interfaces, how do professionalskeep track of what happened and why?Thefinal outputmustbeaccurate, transparent, secure, and defensible.As AI moves from answering questions to planning, acting, and coordinating work across tools, the system behind the interface matters more than ever.

That question is becoming more urgent as AI changes the speed and scale of professional work. The gap between what is plausible and what is correct, between what is fast and what can be trusted, is widening. In legal, tax, audit, and compliance, that gap is not theoretical. It creates real consequences for clients, organizations, markets, and reputations.

At , we call the standardrequiredfor this work Fiduciary-Grade AI™:AI designed for professionals whose workis subject to regulatory oversightand dutiesof careand whereaccuracy,confidentiality, andaccountabilityare critical. In the agentic era, that standard becomes even more important because AI is no longer only generating answers. It is beginning tohelpplan work, use tools, coordinate steps, and help complete complex tasks.And it is increasingly dependent on many interconnected AI technologies, models, and agents working together to produce a final work product.

That is the standard we are building CoCounselto,as it evolves into anagenticoperating systemforthe professions we serve.

CoCounsel is designed to meet professionals wherever high-stakes work happens. Often, that means starting and completing work directly in CoCounsel. Increasingly, it also means making CoCounsel accessible across the broader ecosystem of tools, models, platforms, and partners our customers already use.

The goal is simple: putFiduciary-GradeAI™at the fingertips of our customers, wherever professional work begins, and ensure that when the work matters, it can be trusted.That is what we mean by anagenticoperating system: not a single agent, chatbot, or workflow tool, but a trusted system that can bring together content, context, tools, models, verification, and human oversight to help professionals move work forward.

Starting Work Is Easy. Finishing It Is What Matters.

AI has made it easier than ever to begin work. A prompt can generate a draft, summarize a document, surfacea possible answer, or move a task forward in seconds.

But in professional environments, starting work is not the hard part. Finishing itaccuratelyis.

This need is especially visible in legal, where professionals must ground their work in trusted sources and stand behind the results.When work carries consequences, when it needs to stand up in a boardroom, before a regulator, in an audit, or in court, what matters is how that work is grounded, verified,validated, and made defensible.

That is the role CoCounsel is designed to play.

Trust in professional AI cannot be treated as a feature added at the end.It has to be built into the system itself.Outputs must be grounded in authoritative, continuously maintained sources. They must be shaped by domainexpertise. They must include mechanisms for validation and transparency, so professionals can understand, verify, and stand behind the results.

This is what separates general-purpose AI fromFiduciary-GradeAI™.

General-purpose AI can be extraordinarily useful for exploration, brainstorming, summarization, and productivity. But high-stakesprofessional work requires more than a plausible answer. It requires systems that understand the domain, apply trusted sources, respect privacy and security obligations, and support the human accountability thatremainsat the center of professional judgment.

We are seeing this first in legal. CoCounsel Legal combines advanced AI with trusted legal content, workflows, andexpertise, including Westlaw, PracticalLawandKeyCite,which can be applied tocustomers’ own documents and workproduct. It is designed not only to help legal professionals move faster, but to help ensure their work can be trusted when it matters most.

As Elizabeth F. Salsedo-Surovov, Director of Knowledge Management and Information Resources at Robinson & Cole LLP, recently said:“We’re really excited to be using [CoCounsel Legal] Deep Research… we are using it in Westlaw through our partnership with , in ouron-premiseAI environment to pull in the Deep Research reports so that people are not doing research in theChatGPTsof the world, they’re doing research in Westlaw. …Utilizethe trustedresourceand tools of Westlaw to do your legal research.Don’trely on other AI tools.”

That customer perspective captures the shift underway. Professionals are not just asking for faster AI. They are asking for AI they can rely on.

Building Across the Customer’s Ecosystem

The next generation of CoCounsel is being built for that reality.

CoCounsel will continue to be a primary destination for professional work. For many customers and many workflows, the best experience will begin inside CoCounsel itself, where can bring together content, workflow, context, AI, and various 3rdparty applications and agentsin one trusted environment.

But professional work also starts in many other places. It startsindocuments, emails, enterprise systems, matter management tools, tax workflows, accounting platforms, customer environments, and increasingly general-purpose AI assistants.That is why we are building CoCounsel tooperateacross a broader ecosystem as well.

Through APIs, connectors, Model Context Protocol capabilities, and emerging interoperability standards, CoCounsel can be integrated with and invoked from the environments where customers are already working. In legal, we are beginning todemonstratethis through our work with Anthropic and Claude, enabling customers to connect Claude with CoCounsel Legal so that work can move between a general-purpose AI environment and trusted legal AI system.

And this is just one part of the broader strategy.

We are engaging across the AI and enterprise technology ecosystem,with Google, Microsoft, OpenAI, AWS, Anthropic, and other partners, to make CoCounsel available where our customers choose to work.Through MCP and other integrations, CoCounsel can be accessed from the environments where customers are already working.

This is not about turning into a passive content layer for someone else’s interface. It is about making CoCounsel available as the trusted professional AI system that grounds,validates, and completes high-stakes work, regardless of where that work begins.

The same strategy extends beyondlegal. As CoCounsel continues to expand across , including tax, accounting, audit, risk, and compliance, we are building toward a future where fiduciary-grade AI is available across the professional workflows our customers rely on every day.

What Comes Next

AI is raising expectations for speed and scale across every professional industry. But as the volume of AI-assisted work increases, so does the pressure on accuracy, judgment, and accountability.

The systems that matter most will not be the ones that simply help professionals startworkfaster. They will be the ones professionals can rely on to finish work correctly.

That is the future we are buildingtoward withCoCounsel.

We will continue to make CoCounsel a best-in-class destination for professional work. And we will continue to extend it across the ecosystems where our customers are already working, across models, platforms, partners, and workflows.

Because the future of professional AI will not be defined only by who can generate the fastest answer.

It will be defined by which systems professionals can verify, trust, and stand behind.

That is the fiduciary-grade standard we are building into CoCounsel across .

Share