Connect with us

Accounting

Xero’s JAX said to tame gen AI hallucinations for acconting tasks

Published

on

Small business accounting platform Xero announced that it is beta testing a new generative AI assistant called Just Ask Xero, or JAX, which sports a control system that Diya Jolly, chief product and technology officer, said ensures accuracy and security. 

Speaking during Xero’s annual Xerocon event in Nashville, she noted that Xero is “no stranger to AI” as “it powers a range of our products,” but what’s different here is that JAX uses generative AI to automate tasks and provide guidance through a plain language interface. So while, before, someone might press a button that says “create an invoice,” then type in the line items and then type in the prices and then check the total, users would be able to simply tell JAX to create an invoice, and the AI will pull from the relevant data to deliver the result. 

“All of that is already in your email. You already typed it out once. Why do you need to type it again,” she said in a later interview, noting that it’s “just more natural” to interact with a plain language interface versus navigating through tabs and menus to get things done. 

Jolly said that accuracy is one of the key differentiators for its AI system. The tendency for large language models to give inaccurate information, particularly where numbers are concerned, is well known at this point. This has led to a certain degree of hesitation from professionals to deploy generative AI for serious accounting work (see previous story). Jolly nodded to these concerns, noting that “most of our competitors” are pursuing models that are very generic and prone to hallucinations.

“While there is power in generative AI, it has to be bound for accounting. … We cannot launch something in accounting where we do not have a high level of belief in its accuracy. This is our product. What are we doing if we’re not accurate?” she said. 

To this end, JAX was trained on a very specific set of data. More generic models such as those developed by Microsoft or Google are trained on massive data sets because it is intended for users to apply them to a wide set of functions. Jolly said that JAX was trained on more specialized data, such as being able to recognize an invoice or a quote, or understand terms like cash outstanding or accounts receivable. This helps the AI stay on task and avoid some of the confusion that can come from other models. 

Beyond this, however, the accuracy of the outputs are further bolstered by the fact that JAX was described as a hybrid AI that combines a large language model with machine learning and deep learning models. JAX itself does not actually do the work but, rather, acts as a go-between with the human user and the other AI models. 

So, if a user asked JAX for a cash flow projection over the next quarter, JAX would understand the request; then, it would convert this request to actual machine code which then gets passed onto the deep learning and machine learning AIs on Xero’s servers; these models would then perform the necessary calculations using the data they are allowed to access; the results, in machine code, would then be passed back to JAX, which would then translate the information back into plain language for the user to see. This is all part of what Xero called “JAX Assure” which Jolly described as a sort of control center that keeps the results accurate. 

“Because this is accounting, we want to be a lot more precise. So we can’t leave it up to the generative AI models to tell you cash outstanding. So then we use the machine learning, deep learning models to do the task. We are pretty confident, then, that we’re not going to get hallucinations… because, again, the AI models convert the language but the actual calculations happen with our [other] models,” she said. 

She also highlighted the AI’s mobile compatibility. People can access JAX through a mobile device, so they’re not tied to a desk, they can do what they need to do wherever they are. Jolly said she was often frustrated by the fact that she would go to meetings with “all these bills and receipts” but couldn’t do anything with them until she could get to her computer later. 

“So the fact that I just sent a quote or just created an invoice… the fact you can do it from email, you can do it from WhatsApp, it is extremely liberating and efficient for small business users as well as an accountant. So, being able to get paid, being able to make sure you’re staying on top of what you need to do to get your business moving, I think is cool, because believe it or not most of our businesses, when they have to send invoices or whatever at night, they forget,” she said. 

These features are only the beginning. Jolly, during her presentation, said that JAX, over time, will be in more and more of the Xero platform where it might be able to do things like check for anomalies or find specific types of transactions. Regardless of what it does, though, Jolly said the key differentiator will be its accuracy. 

“I think our accuracy will be our long sustaining [differentiator], like ‘hey we found a way to do gen AI that is accurate. And private,” she said. 

JAX is currently in beta. Those who are interested in taking part can click here.

Continue Reading

Accounting

IAASB tweaks standards on working with outside experts

Published

on

The International Auditing and Assurance Standards Board is proposing to tailor some of its standards to align with recent additions to the International Ethics Standards Board for Accountants’ International Code of Ethics for Professional Accountants when it comes to using the work of an external expert.

The proposed narrow-scope amendments involve minor changes to several IAASB standards:

  • ISA 620, Using the Work of an Auditor’s Expert;
  • ISRE 2400 (Revised), Engagements to Review Historical Financial Statements;
  • ISAE 3000 (Revised), Assurance Engagements Other than Audits or Reviews of Historical Financial Information;
  • ISRS 4400 (Revised), Agreed-upon Procedures Engagements.

The IAASB is asking for comments via a digital response template that can be found on the IAASB website by July 24, 2025.

In December 2023, the IESBA approved an exposure draft for proposed revisions to the IESBA’s Code of Ethics related to using the work of an external expert. The proposals included three new sections to the Code of Ethics, including provisions for professional accountants in public practice; professional accountants in business and sustainability assurance practitioners. The IESBA approved the provisions on using the work of an external expert at its December 2024 meeting, establishing an ethical framework to guide accountants and sustainability assurance practitioners in evaluating whether an external expert has the necessary competence, capabilities and objectivity to use their work, as well as provisions on applying the Ethics Code’s conceptual framework when using the work of an outside expert.  

Continue Reading

Accounting

Tariffs will hit low-income Americans harder than richest, report says

Published

on

President Donald Trump’s tariffs would effectively cause a tax increase for low-income families that is more than three times higher than what wealthier Americans would pay, according to an analysis from the Institute on Taxation and Economic Policy.

The report from the progressive think tank outlined the outcomes for Americans of all backgrounds if the tariffs currently in effect remain in place next year. Those making $28,600 or less would have to spend 6.2% more of their income due to higher prices, while the richest Americans with income of at least $914,900 are expected to spend 1.7% more. Middle-income families making between $55,100 and $94,100 would pay 5% more of their earnings. 

Trump has imposed the steepest U.S. duties in more than a century, including a 145% tariff on many products from China, a 25% rate on most imports from Canada and Mexico, duties on some sectors such as steel and aluminum and a baseline 10% tariff on the rest of the country’s trading partners. He suspended higher, customized tariffs on most countries for 90 days.

Economists have warned that costs from tariff increases would ultimately be passed on to U.S. consumers. And while prices will rise for everyone, lower-income families are expected to lose a larger portion of their budgets because they tend to spend more of their earnings on goods, including food and other necessities, compared to wealthier individuals.

Food prices could rise by 2.6% in the short run due to tariffs, according to an estimate from the Yale Budget Lab. Among all goods impacted, consumers are expected to face the steepest price hikes for clothing at 64%, the report showed. 

The Yale Budget Lab projected that the tariffs would result in a loss of $4,700 a year on average for American households.

Continue Reading

Accounting

At Schellman, AI reshapes a firm’s staffing needs

Published

on

Artificial intelligence is just getting started in the accounting world, but it is already helping firms like technology specialist Schellman do more things with fewer people, allowing the firm to scale back hiring and reduce headcount in certain areas through natural attrition. 

Schellman CEO Avani Desai said there have definitely been some shifts in headcount at the Top 100 Firm, though she stressed it was nothing dramatic, as it mostly reflects natural attrition combined with being more selective with hiring. She said the firm has already made an internal decision to not reduce headcount in force, as that just indicates they didn’t hire properly the first time. 

“It hasn’t been about reducing roles but evolving how we do work, so there wasn’t one specific date where we ‘started’ the reduction. It’s been more case by case. We’ve held back on refilling certain roles when we saw opportunities to streamline, especially with the use of new technologies like AI,” she said. 

One area where the firm has found such opportunities has been in the testing of certain cybersecurity controls, particularly within the SOC framework. The firm examined all the controls it tests on the service side and asked which ones require human judgment or deep expertise. The answer was a lot of them. But for the ones that don’t, AI algorithms have been able to significantly lighten the load. 

“[If] we don’t refill a role, it’s because the need actually has changed, or the process has improved so significantly [that] the workload is lighter or shared across the smarter system. So that’s what’s happening,” said Desai. 

Outside of client services like SOC control testing and reporting, the firm has found efficiencies in administrative functions as well as certain internal operational processes. On the latter point, Desai noted that Schellman’s engineers, including the chief information officer, have been using AI to help develop code, which means they’re not relying as much on outside expertise on the internal service delivery side of things. There are still people in the development process, but their roles are changing: They’re writing less code, and doing more reviewing of code before it gets pushed into production, saving time and creating efficiencies. 

“The best way for me to say this is, to us, this has been intentional. We paused hiring in a few areas where we saw overlaps, where technology was really working,” said Desai.

However, even in an age awash with AI, Schellman acknowledges there are certain jobs that need a human, at least for now. For example, the firm does assessments for the FedRAMP program, which is needed for cloud service providers to contract with certain government agencies. These assessments, even in the most stable of times, can be long and complex engagements, to say nothing of the less predictable nature of the current government. As such, it does not make as much sense to reduce human staff in this area. 

“The way it is right now for us to do FedRAMP engagements, it’s a very manual process. There’s a lot of back and forth between us and a third party, the government, and we don’t see a lot of overall application or technology help… We’re in the federal space and you can imagine, [with] what’s going on right now, there’s a big changing market condition for clients and their pricing pressure,” said Desai. 

As Schellman reduces staff levels in some places, it is increasing them in others. Desai said the firm is actively hiring in certain areas. In particular, it’s adding staff in technical cybersecurity (e.g., penetration testers), the aforementioned FedRAMP engagements, AI assessment (in line with recently becoming an ISO 42001 certification body) and in some client-facing roles like marketing and sales. 

“So, to me, this isn’t about doing more with less … It’s about doing more of the right things with the right people,” said Desai. 

While these moves have resulted in savings, she said that was never really the point, so whatever the firm has saved from staffing efficiencies it has reinvested in its tech stack to build its service line further. When asked for an example, she said the firm would like to focus more on penetration testing by building a SaaS tool for it. While Schellman has a proof of concept developed, she noted it would take a lot of money and time to deploy a full solution — both of which the firm now has more of because of its efficiency moves. 

“What is the ‘why’ behind these decisions? The ‘why’ for us isn’t what I think you traditionally see, which is ‘We need to get profitability high. We need to have less people do more things.’ That’s not what it is like,” said Desai. “I want to be able to focus on quality. And the only way I think I can focus on quality is if my people are not focusing on things that don’t matter … I feel like I’m in a much better place because the smart people that I’ve hired are working on the riskiest and most complicated things.”

Continue Reading

Trending