Read Intuit press releases going back to 2007, from earnings announcements to investor event news. Filter by year or show financial releases only.

Month-End Close on autopilot: write your procedures before the AI arrives. Intuit just showed an AI agent that prepares the monthly close and waits for your approval. The catch: it only works as well as your written procedures. Month-End Close Bookkeeping Audit Log 4-Step Plan Every month the same scramble happens: bank feeds to check, receipts to match, invoices to chase, and a pile of numbers that must be right before you can trust them. This week, one of the biggest names in small business accounting showed what it looks like when an AI agent does the first pass for you. The news: an AI agent for the monthly close. At its Intuit Connect event, Intuit announced Agentic Books Close, an add-on for accounting firms using Intuit Accountant Suite. According to the company's October 27, 2026 announcement, it runs more of the monthly close automatically, following a firm's own standard operating procedures and recording every action in an audit log, while accountants supervise, review, approve and manage exceptions. It is currently a pilot, with a paid add-on planned later. monthly close, prepared by an agent of actions tracked in an audit log entries final until a human approves Intuit says the agent posts pending bank transactions, reconciles accounts and flags exceptions such as possible prepaid expenses for review. Approved items post right away. Note that this is a vendor announcement and a pilot, so results will vary. The point for you is the design, not the product name. The real lesson The agent follows your procedures. If they live only in someone's head, there is nothing for any AI to follow. One attendee summed it up well: write down your procedures, clean up your workflows, and the businesses with clear playbooks benefit first. Why this matters beyond accountants. You may never use Intuit's firm tools, but the pattern will show up everywhere: the AI does the repetitive prep, a person approves, and a log shows what happened. It is the same approval-first idea behind AI guardrails and agent oversight and the reason AI bookkeeping is moving from simple categorizing to running whole routines. Owners who have already captured their know-how in writing are the ones ready to use it. A 4-Step Plan to get your Books agent-ready. * 1Write your close checklist. List every step you or your bookkeeper take each month, in order: reconcile bank and card accounts, review unpaid invoices, record depreciation or prepaid items, send reports. Plain bullets are fine. * 2Name the owner and the approver. Someone prepares, someone different signs off. Decide which exceptions (large payments, unusual vendors) always come to you. * 3Clean the inputs. Tidy your chart of accounts, connect bank feeds, remove duplicate customers and vendors. An agent working from messy data produces confident-looking mess. See payment operations for the cash side. * 4Pilot one month with the log on. Let AI prepare, then compare against your own work. Track hours saved and every correction you made, and fold the corrections back into the checklist. Common mistakes to avoid. * Skipping the approval step because the first month looked right. * Giving broad access when read-only would do for the test. * Automating a broken process. Fix the checklist first, then speed it up. The takeaway. The month-end close is a preview of how AI will handle most back-office routines: it prepares, you decide. The bottleneck is rarely the technology. It is whether your procedures are written down and your data is clean. Do that homework now and any agent, from any vendor, will work better for you. Where gohuman AI fits in. You do not need to wait for a pilot program. GoHuman AI builds AI business automations around your existing procedures, connects your systems in one unified AI tech stack, and keeps a human in the loop for every approval. Book a free discovery call and GoHuman will help you turn your month-end routine into an agent-ready checklist. Related Reading

From prompt engineers to tool builders: Intuit's Vignesh Subrahmaniam on building AI that acts. At TechSparks 2026, Intuit's Principal AI Scientist explained why reliable tools, high-quality data, and consent matter more than ever as AI moves from answering to acting. For the past few years, getting better results from AI meant learning to write better prompts. That is changing fast, as AI systems begin to reason, make decisions, and act across entire workflows on their own. At TechSparks 2026, YourStory's flagship event, Dr Vignesh Subrahmaniam, Principal AI Scientist at Intuit, spoke with Adith Charlie, Managing Editor of YourStory and The CapTable, in a fireside chat titled 'Beyond the prompt: Building AI that thinks, acts, learns'. Drawing on Subrahmaniam's experience of turning AI research into scalable systems, the two explored what builders need to rethink as intelligence becomes increasingly agentic. Teaching machines the game of business. The session opened with a question for Subrahmaniam about what a typical day looks like for him. He was also asked to simplify the concepts of fintech and AI for the audience. He responded with a chess analogy, posing a question: if an algorithm were never taught the rules of chess but was shown a million completed games and their winners, could it deduce both the rules and the strategies most likely to lead to victory? His team at Intuit applies a similar approach to supporting small businesses, as the company serves both consumers and accountants. "How do we learn the game of business by looking at data, by looking at actual actions and outcomes that businesses are taking," he said. The goal is to understand what helps a business win customers, grow quickly, and improve its cash flow. From prompts to tools. The conversation then turned to the shift builders face, from years of learning to write smarter prompts to designing agents that are handed a goal and decide the steps themselves. Subrahmaniam placed prompt engineering in the long history of programming languages, from binary and assembly to Fortran, C, and Python. Large language models let anyone program using plain English. However, this method has limits. Researchers quickly understood that intelligence should not only be in the model. The model becomes an orchestrator, breaking a complex task into a series of actions and tool calls. The real intelligence moves into the tools it calls. Those tools must be built differently too, "not so that a human being can use them, or a software developer can use them, or anybody can use them, but in a way that an LLM orchestrator can really use them". That means high-fidelity tools with highly deterministic output. For builders, the craft shifts from clever wording to dependable components that a model can call on again and again without surprises. The next frontier is beyond text. On where new data will come from, Subrahmaniam was candid that learning from text is now largely a solved problem. The harder challenge is learning from multimodal sources and real-world actions. He pointed to how a small child learns mostly from what she sees, hears and observes, not from reading, and argued that bringing such real-world signals into AI systems is what will help models go beyond generating text to suggesting actions. That, he said, is the next frontier. Data quality still decides everything. Looking ahead at the challenges for a regulated industry, he took the audience back 18 years, to one of the first machine learning labs in Bangalore, where he worked on logistic regressions and multi-layer perceptrons. The biggest problem then was not the algorithm. "Actually, it was data quality," he recalled. That has not changed. Whether a team trains a simple regression or fine-tunes a transformer, "garbage in, garbage out" still applies. Enterprises that can collect, collate, and store data at the highest quality, he said, will have the edge. Consent and privacy from the floor. In the short audience Q&A, an attendee asked how restrictions on where financial data can be stored affect Intuit's work. Subrahmaniam called consent the heart of any learning system and cited the Framingham study, where residents of a Massachusetts town allowed doctors to follow their health over time, which shaped modern understanding of cardiovascular disease. Companies, he added, must also explain the benefits of data use to the wider community. What builders can take away. As the conversation drew to a close, one thread ran through every answer: as AI moves from responding to acting, the foundations matter more than the interface. Prompts will remain a starting point, but the real work now lies in building dependable tools that agents can call on, feeding systems high-quality data, and earning the consent that lets them learn from it. For builders looking beyond the prompt, that is where the next wave of AI will be won.