Editor's note: This piece is adapted from "AI Experiments to AI-Powered Practice," a webinar hosted by Gusto and Karbon. Watch the full session on demand here.
Every month, this report cost Ben Curtis 20 hours.
By the time it was finished, it was already outdated.
He needed it to understand how his firm was performing. Realization, client profitability, team performance. But getting there meant exporting data from Gusto and Karbon separately, stitching it together manually, and rebuilding it from scratch.
So he only did it once a month. Which meant he was always looking backward.
Ben runs Good Measure, a 12-person cloud accounting firm based in Knoxville, Tennessee. His team manages fractional accounting, advisory, and payroll for clients across the country, using Karbon for workflows and Gusto for payroll.
Like most firms, they had already started experimenting with AI. ChatGPT helped with drafting, research, and meeting summaries. Useful, but the way the firm ran had not really changed.
The 20-hour report was still a 20-hour report.
Start with the work that hurts most
What changed was not a new tool or a big rollout.
It was a decision to start with the most painful, most frequent task and fix that first.
He opened Claude and described what he wanted in plain English. What data he had, what he needed to see, and what the output should look like.
"I'm not a coder," he said.
Claude generated a Google Apps Script that pulled data from Gusto and Karbon, combined it, and updated automatically. Ben had to learn what an API key was and where to find it, but that was about it.
The first version did not work. Data came in wrong, formulas broke, and the team did not love it.
"The dashboard sucked at first," Ben said. "We kept going anyway."
That persistence mattered more than the tool itself.
Today, that report updates every 24 hours. Everyone on the team has access - clients, realization, performance - in real time. What used to take 20 hours a month now runs in the background, and the firm can see what is happening as it happens.
Those hours did not just get absorbed elsewhere. They went toward the work that is harder to automate - client conversations, advisory questions, the judgment calls clients are actually paying for.
The work before the work
Before any of that was possible, Ben had to get two things right.
The first was data security. Accounting is a trust-based business. Good Measure has spent years building its reputation, and protecting client data is part of that. So instead of starting with client work, Ben started with his own firm's data. Testing AI workflows internally made it easier to understand how information moved between systems and where he felt comfortable using it.
"Everyone has to make their own decisions," he said. "But we are seeing the most benefit doing things with our own firm internally."
The second was documentation. Good Measure keeps SOPs, client workflows, and internal operating principles in Notion and Karbon. That documentation gives AI the context it needs to produce outputs that reflect how the firm actually works. Without it, AI is generic. For Ben, that foundation made the difference between experimenting and building something the team could actually use.
The part that isn't about AI
Not everyone on the team approaches AI the same way, and that is expected.
At Good Measure, a smaller group of team members who were already excited about AI became the pilot group. Their client responsibilities were reduced by 5 to 10 percent so they had real time to experiment. They test workflows, share what works, and just as importantly, what does not. Adoption is easier when it comes from peers, not just leadership or external tools.
AI has not replaced judgment at Good Measure. Every workflow still has a person reviewing the output. A team member recently caught errors in a reconciliation process because he was paying attention to the AI output. That accountability has not changed. AI handles the repetitive work and cuts review time significantly, often by 50 to 80 percent, but the team still owns the outcome.
"We're not at a point as a firm where a human in the loop doesn't exist," Ben said. "I don't know if that ever happens."
Ben is clear about where his firm stands: still figuring things out. The dashboard is not perfect. Some experiments did not go anywhere. But the direction is clear.
What Ben is building at Good Measure is not unique to his firm. It is a preview of how the firms that will define accounting over the next decade are being built - not by waiting for AI to be perfect, but by starting where it hurts and pushing through the messy middle.
"I think we're going to be better positioned six months from now and two years from now and five years from now because of this," Ben said.
Fixing the 20-hour report is what made it real.
See how to build this in your firm
Ben built his dashboard by hand - learning API keys, writing scripts, manually connecting data from two systems that were not designed to talk to each other. That work was worth doing. It also should not have to be that hard.
The Gusto and Karbon integration helps to remove some of those manual steps. The data that used to require a custom build to connect now flows between these two platforms (and others) automatically. The integration reduces some of the friction of getting started - but the bigger unlock, as Ben found, is deciding which problem to solve first.
If you are still building reports manually, you are not behind. You are just earlier in the process.
Watch the full conversation - AI Experiments to AI-Powered Practice - on demand here.
Join the Gusto and Karbon integration beta here.
Explore Gusto Pro for your firm here.
About Karbon:
Karbon, a global leader in AI-powered practice management software for accounting firms, provides an award-winning, collaborative cloud platform focused on streamlining work and communications within a firm and its clients.
