Dext launches AI Assist bookkeeping tool in North America
Dext has launched its AI Assist bookkeeping agent in the US and Canada for accountants, bookkeepers and finance teams.
Built into Dext's existing platform, the tool learns how users classify transactions, structure data and edit records. It then suggests how to apply those decisions to similar work in future, while leaving final approval with the user.
The launch marks Dext's move beyond document capture and data extraction into automating judgment-led bookkeeping tasks. These tasks often vary across firms, clients and internal finance teams, especially in transaction categorisation, description fields, reporting structures and tax treatment.
That makes them harder to automate with standard rule-based systems. AI Assist is designed to reflect the way each organisation works, rather than forcing teams to reshape processes around fixed automation rules.

Next stage
Dext says its platform has already processed more than 350 million documents, removing millions of hours of manual data-entry work. AI Assist is intended to handle the next layer of bookkeeping by learning from repeated user decisions and suggesting how those decisions should be applied again.
Every suggestion made by the system is visible and reviewable before use. Users can refine or reject suggestions, and that feedback is fed back into the model's understanding of how the organisation works.
Dext is positioning the product around growing pressure on accounting and finance teams. As transaction volumes rise and compliance demands become more complex, firms are being asked to deliver financial information faster while maintaining oversight and consistency.
In Canada, the company pointed to weaker productivity growth relative to long-term trends as one reason businesses are seeking tools that remove operational bottlenecks without giving up control. The US release makes a similar case, focusing on rising workloads across the profession.
Sabby Gill, Chief Executive Officer, Dext, said, "Dext has already helped remove millions of hours of manual data entry from bookkeeping by automating document capture and processing at scale. With AI Assist, we're taking the next step: helping firms and finance teams apply their own judgement and way of working more consistently across every transaction.
"Every organisation, whether a firm or a finance team, has its own approach to categorising transactions, structuring data and managing workflows. AI Assist learns how you work and helps align those decisions automatically, while keeping you fully in control. This frees up teams to focus on higher-value work like insight, advisory and strategic decision-making," Gill added.
Usage data
Dext also disclosed new figures on platform usage. In January alone, users processed 31.4 million receipts and invoices globally.
Based on an estimate of three to four minutes to process each document manually, that volume would have represented more than two million hours of administrative work. Dext says customers spent about 206,000 hours processing those same documents through its platform, implying a reduction in processing time of more than 90%.
Those figures support Dext's argument that bookkeeping software is moving beyond basic extraction to handle more context-dependent decisions. Rather than replacing staff, the software is intended to reduce repetitive review work and leave professionals to manage exceptions, analysis and client-facing tasks.
Stephen Edginton, Chief Product & Technology Officer, Dext, said, "We designed Dext AI Assist to work alongside users, not replace them. The platform already processes financial data at scale with high accuracy; AI Assist builds on that by learning how each team works and applying those decisions consistently. Human judgment remains central to the process, while AI helps reduce the manual effort that slows teams down."
Early feedback
Beta users reported improvements in efficiency and consistency during testing. One example came from faascloud, where the software was used in client work involving manufacturing job codes tied to revenue and cost of goods sold.
"What stands out is how quickly AI Assist becomes useful once you start using it in real workflows. For certain clients and use cases, it's already making a real difference. We're focused on growing without adding more headcount, so using technology to do more with the same team is critical. AI Assist supports that by helping automate parts of the process that would otherwise take time and manual effort," Michael Shafman, Co-CEO, faascloud inc, said.
"For example, we have a manufacturing client where job codes are essential to how we manage revenue and cost of goods sold. AI Assist is able to recognise that context, extract the right information and apply it consistently, which has been a big help. As you spend more time with it, you learn how to tailor it to your workflows and get more out of it. It's easy to see this becoming a standard part of how we work," Shafman added.
The product is rolling out across Dext's platform, which serves more than 12,000 accounting and bookkeeping firms and 700,000 businesses worldwide.