No code, no technical background: This firm owner is powering AI with connected data from Karbon and Gusto

Ben Curtis will tell you he is just a guy. He doesn’t write code and has no technical background. And yet, in the last nine months, he built a firm-wide dashboard his team uses daily, automated his most painful reconciliation process, and cut review time on large cleanups by more than half.

Ben runs Good Measure, a cloud accounting services firm based in Knoxville, Tennessee, and for years his experience with AI looked the same as most firm owners: occasionally useful, but not actually changing how the work got done.

That changed when the data started talking to each other.

The real reason AI experiments stall

Karbon's State of AI in Accounting 2026 Report, which surveyed 593 accounting professionals across six continents, found that 98% of accounting firms are now using AI. 

But there’s a difference between using AI and using AI, and that difference is in the data that you feed your chosen AI tools.

AI is a tool for synthesis, pattern recognition, and execution. But it can only work with what it can see. So when your payroll platform lives in one system, your workflow management in another, and your client records somewhere else, you are not giving AI a reliable source of data to work with. 

Instead, you are giving it fragments of the picture and expecting it to paint a masterpiece.

What Ben built, and why it worked

For Ben, the turning point was a dashboard.

Every month, he was spending around 20 hours manually exporting data from various tools, aggregating it, and running pivot tables to get a clear picture of his firm's operations: profitability by client, realization rates, the story behind how each engagement was tracking. 

It was important work, but it was also unsustainable.

The data he needed was already there. It lived in Karbon, where his team managed all client workflows and tracked time. And it lived in Gusto, where payroll ran. The problem was that no one had connected it.

The Karbon and Gusto integration changed that. With payroll data from Gusto and workflow data from Karbon now flowing into the same connected layer, Ben had something he didn’t have before: a complete, contextual picture of his firm that AI could actually work with.

Using Claude and Google App Scripts, Ben described his desired end state in plain language and let the model produce the code. The result was a dashboard that pulls from both Karbon and Gusto, updates every 24 hours, and gives every member of his team a live read on their clients and their capacity.

What used to take 20 hours a month now runs on its own.

What the Karbon and Gusto integration actually changes

Karbon and Gusto launched their integration in April 2026, connecting practice management and payroll workflows for accounting firms serving small businesses across the US.

For firms already doing what Ben is doing, the integration removes the manual step between the two systems. The data that used to live in separate exports, updated on different schedules and synced only at the contact level, now flows between Karbon and Gusto automatically.

What connected data and AI look like in practice

Good Measure serves a number of church clients. Churches have specific payroll complexities including housing allowances, dual tax status for clergy, and wage calculations that do not follow standard patterns. 

When a new church client comes on board, the onboarding often surfaces months or years of unreconciled payroll liabilities and uncategorized transactions.

That used to mean a lot of manual, painful cleanup work.

Now, Ben's team pulls the payroll data from Gusto alongside the accounting records in Karbon and feeds both into purpose-built AI workflows. The AI assists with the analysis, and the team reviews the output and handles what requires judgment. What used to take days now takes hours.

When payroll data and practice management data are in the same connected layer, AI can actually do something useful with both of them. When they are not, you are back to exporting, aggregating, and hoping nothing falls through the gaps.

How Ben got started with AI

Ben's approach was deliberate. He started with his own firm's data, not client data. Getting comfortable with how the tools worked, what AI could see, and how data was flowing between systems gave him the confidence to extend it to client work. Testing on your own firm first is a low-risk way to build that understanding.

The first process he chose to tackle was the most painful one, not necessarily the most interesting or the one with the best theoretical ROI. 

Ben understands that using AI is more than about firm growth; it’s also a people strategy. So he started with the process that his team found the most frustrating to do manually. That is usually where the highest return actually lives.

One of the keys to success was documentation. Good Measure has always kept detailed records of its workflows, its client profiles, and its internal operating standards. That documentation became the context that made the AI genuinely useful rather than just functional.

None of this required a technical background. The skills built through working with AI, including testing and refining the output, he believes, will benefit him for a long time.

A practical question to ask yourself

Twyla Verhelst, CPA, VP of Industry Relations and Community at Karbon and a practitioner herself, made the point during Karbon’s recent Claude Training for Accountants webinar: the firms that are pulling ahead are not waiting for a perfect moment. They are getting started, iterating, and building the muscles that will matter over the next decade.

Ben agrees. He doesn’t think Good Measure is there yet. He is still building. But the gap between a firm that has connected its operational and payroll data and one that has not is widening, and the difference is not abstract.

It’s 20 hours a month saved. It’s a payroll reconciliation that used to take days. It is a dashboard that updates while you sleep.

AI did not create those outcomes. Connected data did. AI made them faster.

The question to ask yourself is a simple one: what does your AI actually have visibility into right now, and what is it missing?