About Scale

Scale AI is one of the world’s leading AI infrastructure companies, powering the data pipelines, model evaluation systems, and production workflows used by advanced AI teams in both industry and government.

The company operates at exceptional scale and complexity, processing billions of data points across a global footprint. Its finance and accounting teams support large transaction volumes, multi-entity structures, and rigorous audit and reporting needs, making it a compelling proving ground for agentic close automation.

Scaling accounting at the speed of Scale

As Scale grew, its accounting operation grew dramatically more complex with it: multiple entities, multiple currencies, a global contributor workforce, and roughly $500 million in transactions flowing through the books every month from more than ten data sources. Josh Waldron, SVP of Finance and Accounting at Scale AI, has watched that transformation firsthand.

“When I first got here, we were at, call it, $50 million, and we’re well over a billion now. We’ve expanded across the globe, to multiple entities and currencies, with a significant amount of transaction volume flowing.”

The challenge wasn’t simply closing faster. Scale needed an accounting operation capable of keeping pace with the business without adding people every time complexity increased. And Josh had no interest in solving that problem by throwing headcount at it.

“We didn’t want to just invest in a ton of people. That’s not always the long-term solution.”

The bottleneck: 15 days every month just to reconcile cash

When Josh joined Scale, the company was taking at least 30 days to close its books. At times, the team was effectively closing two months at once.

The team first rebuilt the accounting foundation. But as the operation matured, one bottleneck became impossible to ignore: cash. More than ten sources of cash activity flowed through five banks and multiple systems. Every month, accountants had to pull that activity together and reconcile it manually in Excel.

Cash alone consumed 15 days of every month. “Pulling all of the data from different sources just took a tremendous amount of time. Fifteen days of every single month, we just took to go and reconcile all the different transactions.”

Scale already used FloQast to manage its close. But tracking whether accounting work was complete didn’t eliminate the work required to complete it. The team still had to prepare reconciliations, journals, schedules, and supporting evidence manually.  And in a company moving as quickly as Scale, a long close created a second problem: by the time the numbers were ready, the business had already moved on.

“If you’re closing in 20-plus days, by the time you close, the data is almost irrelevant, especially in a hyper-scaling company.”

Choosing an AI-native accounting platform

Josh had spent more than two decades in accounting, and his standard for automation was straightforward: it had to absorb work without creating cleanup. “One of the things that is super important is the accuracy and completeness of the tool. When it automates it, it automates it correctly, and we’re not doing more work cleaning everything up.”

Scale ran a formal evaluation that included FloQast, Numeric, and Maxima. The distinction Josh saw was architectural. Other platforms were incorporating AI into existing close software. Maxima had been designed around AI performing accounting work from the beginning.

“They had more bolt-on AI, versus at its core leveraging AI like Maxima. We knew Maxima could get to where FloQast was. We saw there was a lot more functionality and capability versus someone that started out as a software tool and is now trying to catch up.”

But AI alone wasn’t enough. Scale needed direct access to the systems where transactions originated, controls and audit trails that could withstand scrutiny, and a platform capable of expanding across the close rather than automating a single workflow.

As Josh put it: “I don’t want someone who comes to me and says, I have a close tool that will track how you’re doing on close. I want something that’s going to do my cash recon, accrual recon, close automation, all those types of things in one stop.”

From a 15-day cash reconciliation to a continuous close

Scale started where the pain was greatest: cash.

Maxima connected cash activity across five bank accounts and five countries directly with NetSuite, bringing Scale’s disparate transaction sources into one accounting workflow and applying the team’s matching logic automatically. The result was dramatic

A process that had consumed approximately 15 days every month fell to roughly one hour.  

But the more important change was when the work happened. Instead of waiting until month-end to begin reconciling cash, Maxima continuously matches transactions as activity occurs. The accounting team enters close with most of the work already completed and focuses its attention on exceptions.

“It’s more of the continuous close cycle that we were looking for. Every single day it’s matching your transactions, and anything that comes up, we can easily flag, identify, focus on, and resolve very quickly.”

Today, 98% of Scale’s cash transactions are reconciled automatically.

Expanding automation across the close

Once cash proved the model, Scale expanded Maxima into the rest of the accounting operation.

• Recurring journals and balance sheet schedules came next. Prepaids, fixed assets, JPMC cash, prepaid commissions, Deel payroll accruals, Coupa accruals, and ASC 842 lease entries from Visual Lease moved into Maxima. Instead of manually formatting files and preparing uploads, Max prepares the accounting work and supporting detail for review and approval before it reaches NetSuite. 

• Scale then migrated its close management entirely off FloQast, bringing its global entities and close checklists into Maxima. Accountants can see what needs their attention in Close Overview and My Work, while the underlying accounting workflows and evidence live in the same system.

• Flux analysis followed. Previously, flux could consume another two to three days after the books closed. With the underlying transaction context already in Maxima, Max prepares variance explanations as the close progresses. “Now we close, and the next morning we’re able to have our flux meetings because of all the prepared flux that’s already been done by Maxima.”

Scale is now extending that model into business-unit flux and self-serve reporting, AI-generated insights, and a lease accrual and roll-forward agent that works property by property. Across each workflow, the division of work stays consistent: Max prepares the accounting work. Scale’s accountants review the output, investigate exceptions, and exercise judgment.

Automation without giving up control

For Josh, automating accounting work was only valuable if his team could trust what came out the other side. That means knowing where a number came from, how it was calculated, what changed, and who reviewed it.

Every workflow in Maxima maintains transaction-level lineage from source to ledger, preparer-reviewer separation, supporting evidence, and an audit trail of the work performed. “I want to see where everything’s coming from and what’s happening, and gain the confidence that everything is accurate and complete. Then I can go to my CFO and be confident in what I’m providing.”

That human-in-the-loop model is central to how Josh thinks about AI in accounting. He believes agents can ultimately perform roughly 75% of the team’s initial accounting work, while accountants remain responsible for the controls, review, and judgment the function requires. “As our accounting operations have become more complex, we’ve looked for ways to automate more work without compromising accuracy, auditability, or control. We get Maxima’s agentic system of work, and now Max, a 24/7 accountant that can prepare accounting work across our finance stack while operating within our existing SOX requirements, controls, and approval workflows.”

A three-day close and an accounting team that didn’t have to scale with growth

Scale now closes its books in three days, down from seven, with 98% of cash transactions reconciled automatically. At the same time, the accounting team has kept headcount flat through 300% revenue growth.  

That capacity proved especially valuable as the company navigated a major investment from Meta, leadership transitions, and multibillion-dollar customer negotiations. Instead of having the accounting team consumed by the mechanics of close, Josh could redirect its attention as new priorities emerged. “Having the platform in place that eliminates all of the manual work has allowed us to just focus on those things as they’ve come up,” he says. “We were able to make it successfully through all of these different transitions because the team just had the time to do so.”

The shift has also changed how the accounting team works with the rest of the business. With less time spent gathering data and preparing routine close work, accountants can spend more time understanding what is happening in the numbers and helping business leaders act on them. Josh points to decisions around headcount as an example: rather than simply reporting the numbers, his team can work with business partners to understand the request, evaluate the underlying data, and show the expected return.

For Josh, that is where the role of accounting is headed. “You’re not seen as the back-office individuals,” he says. “You’re a true partner to the business.”

The destination: continuous close

There is still considerably more work Josh wants to automate. He estimates that 25–30% of the team’s total workload is automated today, with the potential to reach roughly 75%. Vendor accruals that still require back-and-forth with suppliers, forecasting, and other transactional work are among the next areas his team is looking to automate.  

The longer-term goal is to make the close increasingly continuous. Cash already works this way: transactions are matched throughout the month, exceptions are surfaced as they arise, and accountants enter close with much of the reconciliation already complete. Josh sees the same model eventually extending across more of the books, with accounting work prepared and reviewed as activity happens rather than accumulated for month-end.

It is a significant change from how he has spent much of his 20-year accounting career. “Twenty years ago I would have said there’s no way in this world,” Josh says. “Now I’m a strong believer that in the next few years you press the easy button daily, you get the results of the day, everything’s booked into your general ledger, and you’re driving the business based on what you’re seeing.”