Firm Award · Strategy & Governance
AI Native Accounting® Foundation Case Study
2026 Firm Award for Strategy & Governance Winner: CBIZ, Inc.
Lead: David Fisher, Vice President, Artificial Intelligence
Eleven people. Nine thousand five hundred colleagues. Under three months to full adoption. That is the arithmetic behind CBIZ's award, and it is worth understanding how a team that small moved a firm that large.
When David Fisher joined CBIZ in 2025 he inherited a decent starting position and a much larger opportunity. CBIZ had just absorbed Marcum and grew past Grant Thornton in headcount. Its people had a working AI Solution. What the firm didn't have was AI that could reach into its own accumulated knowledge, or anything close to everyday use across the business.
Fisher had done a version of this before. He came to CBIZ from KPMG, where he led the Global Audit Incubation Hub and built AI products for more than 80,000 professionals across 140 member firms. That role started small, he was the technology champion on a single audit engagement in Washington, D.C., which turned into doing it locally, then nationally, which eventually scaled across KPMG's global member firms.
What follows is what he and a team of eleven did next.
Firm
CBIZ, Inc. national professional services advisor, headquartered in Cleveland, OH
Size
9,500–10,000 team members
Structure
Publicly traded company (note that since the award designation, CBIZ has announced plans to be acquired by Grant Thorton)
Service lines
Accounting, Tax, Advisory, Insurance, Benefits & HR Services and Technology Solutions
Who ran it
David Fisher, Vice President, Artificial Intelligence leading AI inside the firm, and as of July 2026 is also heading up CBIZ's AI consulting services
Team
11 people approximately half product engineers, half transformation specialists
What employees use
Microsoft Copilot, for every team member
Governance
An AI steering council, a firm-wide usage policy, and a set of guiding principles built on the NIST AI Risk Management Framework. ISO/IEC 42001 certification is in progress.
Recognition
The AI Native Accounting® Awards recognize measurable, responsible, replicable progress not innovation theater. Financial Optics won the Small Firm Award for Strategy & Governance because they did something most small firms skip entirely: they built the governance before adopting AI into the practice.
At CBIZ, the AI work never sat in the IT department. It was funded out of a central AI budget, on the reasoning that this was a transformation problem that happened to involve software. The order of operations was unusual too. CBIZ stood up its steering council, usage policy, and guiding principles before buying anything. It built a required training course around them. And only then did licenses start switching on, one person at a time, as each one finished the course. Sixty minutes of AI Fundamentals was the price of admission to Copilot.
That last piece matters, though it's worth being precise about what reached full adoption, because two things did. Every one of CBIZ's 9,500-plus team members completed the AI Fundamentals training. And nearly every one of them adopted Copilot full adoption was the target Fisher set when the rollout began in April, and the firm hit 99% in June. Licenses issued is the number most firms end up reporting, but CBIZ can point to a workforce that is both trained and actually working in the tool.
Background
CBIZ didn't build this from scratch, and the story is more useful because of that.
A lot of the firm's technology group came over in the Marcum acquisition in 2023–2024. Marcum had already built its own private chatbot on Azure OpenAI and called it Ask Marcum. It ran internally for about 18 months and helped carry people through the disruption of the merger.
CBIZ renamed it Vertical Vector and relaunched it inside the firm.
Once Fisher arrived, the team's read was that Vertical Vector had been right for Marcum's size but the combined firm could do more.
So they ran a real search: 13 startups plus every major model provider. Most got ruled out for one of two reasons: the startups either couldn't reach CBIZ's data or were built for one narrow slice of the business. And the big providers' enterprise offerings had nothing to plug into no Outlook, no Excel, no SharePoint.
They picked Microsoft Copilot, and they were honest with themselves about the tradeoff. Copilot may have, at the time, been a bit behind the newest, flashiest models. They took that deal, because Copilot could reach 10 to 15 years of the firm's own accumulated knowledge the documents, emails, files, and history already sitting in their Microsoft environment.
That tradeoff is the one most firms get backwards. They chase the highest-scoring model and end up with something brilliant that knows nothing about their business.
Two phrases guide everything CBIZ builds: "build once, use many" and "bring your own context."
The Work
Fisher is clear about the order. Before anyone was trained, and long before anyone got a license, CBIZ set up:
Then all of it went straight into the training instead of into a folder nobody opens. It kept showing up afterward too, in follow-up sessions, a spotlight series, and internal challenges, with the responsible-use piece called out every time.
CBIZ built it in-house, and by Fisher's own description it's a lot less painful than the average compliance video. It runs in three parts:
The third part is why it landed. Nobody sat through an abstract lecture about AI. They watched their own job get easier. Finish the session and you're certified in AI Fundamentals, and your license switches on.
Level two training is a four-week hybrid program, and it's more selective because CBIZ asks a lot more of the people in it. The first group had 50 people. They learn advanced prompting and how to spot a use case and turn it into something the whole firm can reuse.
The prompting part is smarter than it sounds. The team kept seeing people write enormous prompts full of blanks to fill in clunky, and useless to anyone else (AI slop). So they taught people to have a conversation with the tool instead, and to build something once that many people could then use. Hence "build once, use many."
Champions come out certified, and (this is the part most firms skip) they come out funded. They get charge codes and time codes to run lunch-and-learns, hold office hours, and record short demos. CBIZ pays for that work instead of asking for it as a favor.
CBIZ pairs early-career staff one-on-one with senior leaders, and the junior person does the teaching.
This fixes two things at once. Senior staff who've been avoiding the tool, get to be beginners in private, which is much easier than being a beginner in a room full of colleagues. And junior staff get real face time with firm leadership. In a profession built around seniority, flipping who teaches whom is a genuinely big deal.
Fisher's rule: don't make it a tech issue, make it everyone's opportunity. His account of the budget decision explains why.
There's a wrinkle worth noting here. Not only did AI never sit under IT at CBIZ when the firm's CIO retired recently, the head of transformation Peter Scavuzzo took over the CIO role.
So this isn't really a story about AI versus IT. If anything, it's a story about the transformation side ending up with IT rather than the other way around. IT is built to keep things running, secure, and on budget, which makes it very good at managing systems and not especially suited to changing how thousands of people do their jobs. Those are different skills and they get measured differently.
CBIZ sorts its AI work into four levels, which gives everyone a shared way to talk about what's being built:
Level What it is
All four are built on "bring your own context," and the platform was built for the whole organization rather than one practice group.
Fisher is held to value realization revenue and cost savings. The method is specific and refreshingly unglamorous:
No blended productivity claim. Just a use-case-by-use-case tally that holds up when someone asks how you got there.
Asked point-blank whether any of this replaces headcount, Fisher said no. The skills people need are shifting, but the focus right now is bringing the team along. Then he made the connection that ties the whole program together: the same way the firm's own data was what made the Microsoft decision right, what people bring to the tool is what makes the rest of it work. Same idea, applied to humans.
Outcomes
Measure
Outcome
Copilot adoption
Copilot adoption 99% of 9,500–10,000 team members
Training completion
Training completion 100% AI Fundamentals, required before a license activated
Time to get there
Started April 22, done before July 22
Assistants built by staff
~2,000 in a few months
Champions certified
First group of 50 through the four-week program
Team behind it
11 people
Groups leading the way
Benefits & Insurance ("crushing it"), then Advisory interest was broad, but those two stood out for creativity
Certification
ISO/IEC 42001 in progress; CBIZ has already helped five clients earn it
In closing
It's worth saying why this award category exists at all. Governance is an unglamorous word, and in most technology rollouts it describes the document you write so that someone can point to it later. AI doesn't really work that way. A license doesn't determine what an AI tool does the person using it does, every time they open it. So the judgment about how to use it, and why, and toward what end isn't something that can sit beside the training. It's close to being the same thing as knowing how to use it well.
That's the part of CBIZ's story we found most instructive. The steering council, the usage policy, and the guiding principles were all in place before anyone had access, and then they went straight into the training that unlocked the license. Nobody learned the tool in one session and the rules in another. People got both at once, because at CBIZ the two aren't really separable.
What's more is none of this took a large team, which is maybe the most encouraging part: eleven people reaching 9,500 colleagues. What it did take was deciding early what the firm wanted AI to be for, and then building the governance around that answer rather than around the tool.
That's the distinction we keep coming back to at the Foundation. Being AI-enabled is more or less available off the shelf. Becoming AI-native is more deliberate work, since how and why AI gets used has to be embedded in how the firm operates rather than layered on afterward. What CBIZ showed is that this doesn't have to take long it just has to come first.
CBIZ did not design this program to sell it. The program came first, and the client offer came out of it. The firm now runs the same sequence with clients under a framework it calls IMPACT, starting where CBIZ started, with strategy and governance before technology, responsible-use practices built in from the start, employees enabled through practical training, and outcomes measured against real business results. It is the only pitch CBIZ can make honestly, because the firm did this to itself before it did it for anyone else.
David Fisher is Vice President, Artificial Intelligence at CBIZ, Inc., where he leads the firm's enterprise AI strategy. As of July 2026 he also heads CBIZ's AI consulting services.
Fisher joined CBIZ from KPMG LLP, where he led the Global Audit Incubation Hub and developed AI products for more than 80,000 professionals across 140 member firms.
On joining CBIZ, Fisher said:
