Transactional compliance isn’t running on the calendar it used to. E-invoicing and continuous transaction controls mean regulators want visibility at the pace of the transaction, not months later in a filing. That shift is accelerating faster than most manual processes, or most off-the-shelf AI, can keep up with.
CoCounsel is the AI technology Thomson Reuters is incorporating into its solutions. This article is the first in a series exploring how that technology is being applied across the transactional compliance and tax lifecycle, beginning with the thinking behind it before examining the products and innovations it powers.
Ray Grove, Head of Product for Tax and Trade at Thomson Reuters, leads the product strategy behind these innovations. We spoke with Ray about the principles guiding how Thomson Reuters applies CoCounsel across its solutions, the role of trust and human judgment in AI-assisted work, and why meaningful innovation starts with work that matters.
Q: Where’s the line between what AI can handle and what still requires human judgement?
Q: Where do the answers behind CoCounsel-powered solutions come from?
Q: How are customers responding as more of the work shifts from compliance to judgment?
Q: Why meet customers where they are instead of building something new?
Q: If someone is comparing options right now, what’s the one question they should be asking?
Q: Thomson Reuters describes CoCounsel and the AI capabilities as Fiduciary-Grade AI™. Why give it a distinct name, and what does the term exclude that generic AI tools might allow?
Grove: A lot of that comes down to who our customers are and the kind of work they do. This isn’t a convenience use case. It is high-stakes work with high-stakes risk if you get it wrong, and “good enough” doesn’t clear the bar.
I think about it in four parts:
1. Authoritative content — domain-specific, curated content that professionals actually stake their reputations on, not web-scraped data.
2. Privacy and security safeguards — rigorous data governance; customer data is not used to train third-party models, and customer information remains isolated to that customer’s environment.
3. Subject-matter expertise — shaped by domain experts (lawyers, tax professionals, risk specialists) who provide the human judgment that trains and validates AI systems.
4. Transparent, verifiable outputs — designed so professionals can trace every output back to its source and verify it before putting their name on it.
Q: Where did you have to say no — a feature, a shortcut, something the business wanted — because it didn’t clear the bar?
Grove: I told the team to stop shipping small feature updates and basic customer delights and instead put a focused team on a problem we knew mattered. It’s not about doing more things. It’s about doing fewer things that matter more.
That’s how we got ONESOURCE Sales and Use Tax powered by CoCounsel and ONESOURCE Indirect Compliance powered by CoCounsel. Most of what we brought to market was built in 90 days, but speed wasn’t the only measure of success. Customers saw the value, and we saw sales within the first 60 days of launch.
Q: Where’s the line between what AI can handle and what still requires human judgement?
Grove: That line will keep moving as the technology matures, but one thing won’t change: when it comes to the decision or the filing, the human owns it. The human owns the outcome, and the fiduciary responsibility that comes with it.
What we own is getting professionals as close to that decision as quickly and safely as possible. That’s the whole point of being contextual, accurate, reviewable, and expert. We want professionals to have the context they need to understand and validate an outcome before putting their name on it. When additional information or judgment is needed, it brings you into the loop. Then you, the expert, make the call.
Right now, we’re continuing to close that gap. It takes time and it takes interactions. The more a customer works with it, the more context it builds around their business, and the better it gets for them specifically.
Q: Where do the answers behind CoCounsel-powered solutions come from?
Grove: It depends, honestly, and that’s part of what makes this hard to copy. Our tax engines and e-invoicing tools are deterministic. They’re calculators applying rules. What’s different about the reconciliation work is that it pulls context from a much wider set of sources, including e-invoicing, tax determination, a customer’s ERP and other business systems, and their last-mile compliance systems.
We bring all of that together, reconcile against it, and identify errors or anomalies. In some cases, the system repairs them; in others, it gives the customer the chance to approve the repair. That process creates truth in the transaction flow, which creates truth in the data, which creates truth in compliance.
Customers using ONESOURCE Indirect Compliance powered by CoCounsel have reported cutting reconciliation workload by 60%. This is capacity handed back to professionals for judgment work instead of data cleanup. Thomson Reuters customers, including JLL, have said as much publicly.
Q: How are customers responding as more of the work shifts from compliance to judgment?
Grove: They love it, honestly. Tax is one of the most strategic parts of any organization, but it rarely gets the time to help drive outcomes. Compliance isn’t optional, and increasingly it isn’t periodic either. It’s operational. It happens at the pace of every transaction, which makes it a linchpin in whether a sale closes or a shipment goes out.
A lot of the work professionals do today is cleanup: organizing data, wrangling systems, reconciling the mess. None of that is strategic work, and none of it is judgment work. It’s just work.
Q: Why meet customers where they are instead of building something new?
Grove: There are two choices, and it’s something I think about as a one-way door versus a two-way door. It’s a conversation I have in product strategy a lot. When you’re delivering a new capability like this, there’s two ways to approach it. One, you say, “This is such a new experience, I’m going to start over from scratch, go to market, and blow my customers away with something entirely new.” That’s the one-way door. There’s a lot of reasons to do that, but it’s usually not my preferred choice, because customers may not be ready for that level of change that quickly. So, I made the choice of a two-way door instead, bringing the capability into the systems customers already use, rather than asking them to adopt something entirely new. Overtime however, I see the experience evolving into something very different from where they are today. That evolution we want to experience with them building trust along the way.
Q: Tax and compliance data is some of the most sensitive information an organization has. How do CoCounsel-powered solutions protect that data, and how is it used?
Grove: This is genuinely sensitive, privileged data. It’s not just “a transaction happened.” Taken together, it’s a picture of a customer’s accounts, entities, locations, customers, and suppliers that doesn’t exist anywhere else in one place. Keeping that secure has been a cornerstone of our products for a long time.
On training specifically: we don’t train on personal or customer data unless a customer has explicitly agreed to it. Where we build a semantic layer for a customer, it’s partitioned to their own experience, so it makes their product better for them. Customer data remains isolated to that customer’s environment rather than being shared across customers.
Q: If someone is comparing options right now, what’s the one question they should be asking?
Grove: The question I’d want a customer to ask themselves is, “Am I trying to solve a problem, or the problem?” We solve the whole thing, from determination through e-invoicing, reconciliation, and global trade management, and we do it more fully than anyone else in the category. When you think of processes it’s important to consider that trust and reliability matter just as much as capability.
I’d also point to durability. We’ve been in this space long enough, and successfully enough, for organizations to trust us with the work that matters most, and that’s not something you get from speed alone. We move quickly, but we’ve earned the right to be trusted with the hard problems, and that combination is harder to find than people think.
What’s next
Ray’s answers point to something bigger than any single feature. Fiduciary-Grade AI, which is built with a clear line between what the technology decides and what always stays a human call, backed by a security and privacy posture designed for how sensitive this data is. That’s the foundation of every ONESOURCE product built with CoCounsel.
To learn more, request a free demo of ONESOURCE Sales and Use Tax powered by CoCounsel or ONESOURCE Indirect Compliance powered by CoCounsel and try it for yourself.
