With AI now ubiquitous—drafting emails, summarizing meetings and generating images —we must move past asking what it can do. The new, essential question is: what can it reliably and consistently do for you?
That’s a high standard, especially when it comes to money. Financial tasks like filing taxes, managing payroll, reconciling books, and staying compliant isn’t just about automation, but about accuracy and superior results. Yet for millions of small and mid — sized businesses and individuals, these tasks can still take up time, cause stress, and hold back business growth or financial wealth building.
That's the problem Intuit is here to solve.
With approximately 100 million customers and more than four decades at the leading edge of experience in financial technology software, Intuit has long been known for products like QuickBooks, TurboTax, Mailchimp and Credit Karma, and Mailchimp. For more than a decade, Intuit has been driving AI innovation to help our customers solve their most important financial problems. In 2019, the company doubled down on AI—not just as a layer on top of its products, declaring its intention to completely reinvent itself as an AI-driven expert platform that combines artificial intelligence with human expertise to provide "done-for-you" financial solutions for consumers and small businesses
By numbers, Intuit has 625,000 customer and financial attributes per small business, as well as 70,000 tax and financial attributes per consumer, and generates 60 billion machine learning predictions per day. With this robust data set, Intuit is delivering personalized AI-driven experiences to its customers, with speed at scale.
At the heart of this reinvention is Intuit's proprietary Generative AI Operating System (GenOS). Built from the ground up to handle complex financial tasks, GenOS is what scales "done-for-you" AI and agentic AI experiences across Intuit’s platform and products—driving durable growth
We sat down with Jayanth Saimani, Distinguished Software Engineer at Intuit, to unpack why the company built its own “paved road” for responsible development of personalized AI customer experiences on its platform, how it works under the hood, and what it means for the future of financial automation at scale for Intuit.
Q: What prompted Intuit to build GenOS from scratch? Was there a moment where it became clear that existing AI tools weren't enough?
Jayanth: Generic AI models can summarize text, draft emails, and even write code. But when it comes to something as complex and regulated as financial work—from filing taxes to managing payroll—accuracy and trust aren't optional. Off‑the‑shelf commercial AI models and development tools simply couldn't meet that standard.
Intuit's strategy is to build an AI-driven expert platform that combines artificial intelligence and human intelligence (AI+HI) to solve customer problems by delivering "done-for-you" experiences. To bring that vision to life, we had to build our own GenOS from the ground up
It gives us a proprietary foundation to train custom Financial Intuit Large Language Models (LLMs) on decades of financial data sets and deliver personalized experiences, enabling a level of domain‑specific intelligence and compliance that generic models can't match. These state-of-the-art models currently power agentic AI capabilities in QuickBooks Online and Intuit Enterprise Suite, with several agents in development for future product releases.
GenOS provides a unified engine for thousands of Intuit technologists, providing every developer a safe, standardized way to build and scale AI‑driven experiences across our ecosystem. Unlike third‑party AI tools, GenOS was designed to empower Intuit technologists to turn ideas into groundbreaking agentic AI experiences for our customers that connect customers to a virtual team of AI agents and AI-enabled human experts to automate everyday tasks and manage complex workflows and entire processes.
These capabilities make GenOS foundational to Intuit's transformation.
Q: You call GenOS a "Generative AI operating system." Why does that distinction matter?
Jayanth: GenOS isn't just a toolkit. It's the engine behind how we build AI at scale. Calling it an "operating system" reflects what it actually does: it brings together our custom models, responsible AI and governance, developer tools, and AI capabilities into a secure, unified platform. That's what allows teams across Intuit to build intelligent, compliant, and agentic experiences at speed
Q: GenOS combines data, models, developer tools, and security. How do the pieces come together - and how do they help teams go from idea to working agent faster?
Jayanth: GenOS is built as a layered system, with each layer designed to handle a critical piece of the AI development puzzle.
At the core is our secure data and privacy framework—an absolute must when handling financial information. Built on top of that is GenRuntime, our orchestration engine that oversees the execution of AI tasks
Finally, at the heart of GenOS are our Financial Intuit LLMs. These aren't generic models. They're trained on decades of anonymized tax, accounting, and financial data, which gives them a deep understanding of context, compliance, and what we call financial "common sense." Early results with these models demonstrate 5% improved accuracy and 50% reduced latency for some accounting workflows compared to certain general-purpose off-the-shelf LLMs, which makes a huge difference when you're running real workloads at scale.
Then there's GenUX, our design framework that ensures consistent, trustworthy interfaces across all Intuit products.
What brings all of these layers to life is the developer experience. GenOS includes a set of integrated tools—what we call a "paved path" that helps teams go from idea to production-ready agentic experiences without reinventing the wheel.
- GenStudio is the sandbox—this is a place to prototype, experiment with prompts, and validate ideas quickly.
- AI Workbench is the full professional environment for developing and managing production-grade agentic experiences.
- Agent Starter Kit gives teams templates and pre-built components so they can skip the boilerplate and get straight to innovation.
Together, they drastically reduce time-to-value, allowing us to go from idea to live, compliant AI experiences quickly and securely. However, it’s important to note that this stack doesn't just enable faster development; it makes "done-for-you" agentic experiences possible: intelligent, proactive agents that can run payroll, track payments, or automatically follow up on overdue invoices, with seamless access to a human expert (tax, bookkeeping, accounting) whenever a consumer or small/mid-market customer needs it. These aren't features bolted onto existing software. They're autonomous financial workflows that only a system like GenOS can support safely.
Q: GenOS is accessible to Intuit technologists. How has it changed the way your teams build?
Jayanth: It's been a radical shift in how we think about building. Before GenOS, teams often had to start from scratch. Now, instead of asking "What feature should I build?" developers ask, "What's the best AI agent to orchestrate for this task?"
GenOS gives them the foundation to move faster and focus on innovation. It provides a consistent, secure "paved path" complete with tools like the Agent Starter Kit, evaluation services, and built-in guardrails. That means less time spent wiring up infrastructure, and more time crafting high-impact, intelligent workflows. It also offers end-to-end tracing and observability, and dashboards which help measure the quality of agents developed.
And this shift has democratized AI development across Intuit. Whether you're working on QuickBooks, Mailchimp, TurboTax, or Credit Karma, you're not building siloed features anymore, but contributing to a virtual team of agents that can collaborate across products. That's unleashed a wave of innovation we've never seen before.
Q: What's one example of a "done-for-you" experience that shows this in action?
Jayanth: Let’s consider QuickBooks. Intuit’s new AI agents help businesses – at every stage – scale, enabling them to grow on a platform they know and trust. The AI agents complete day-to-day tasks, including managing customer leads, tracking payments, sending invoices, and reconciling a business’s books. In addition, new, integrated collaboration tools allow a business owner and their accountant to seamlessly communicate and work together in QuickBooks. With the automation of workflows, collaboration with trusted experts, and the ability to deliver real-time, personalized, and actionable insights, businesses can move faster, operate smarter, and uncover greater efficiencies and growth
Here are a few examples of done-for-you agentic AI experiences in the hands of Intuit QuickBooks customers today:
- Payments Agent - Optimizes and grows cash flow by getting businesses paid an average of 5 days faster with tailored acceleration strategies that predict late payments, automate invoice tracking, and create and send invoices and reminders.
- Accounting Agent - Automates bookkeeping and transaction categorization, and assists in reconciliation, delivering cleaner, more accurate books. Recently enhanced with expanded anomaly detection, AI-powered reconciliations, and more accurate automated transaction categorizations across classes and dimensions.
- Finance Agent - Provides a comprehensive approach to the financial management of growing, mid-market businesses with robust reporting, KPI analysis, and scenario planning and forecasting based on performance and peer benchmarking, helping finance teams make smart decisions to achieve financial goals.
The new AI agents - embedded in the Intuit platform alongside a Sales Agent, Payroll Agent and Project Management Agent - create a more powerful QuickBooks. A virtual team working on our customers’ behalf, transforming how work gets done with a seamless path to human experts whenever needed
Behind the scenes, Intuit is streamlining end-to-end agent development by bolstering its suite of platform capabilities and tools within GenOS.
Q: Financial data demands a higher standard. How does GenOS deliver safety and compliance at scale?
Jayanth: Overall, Intuit’s AI-driven expert platform and products are built in keeping with the company’s commitment to data privacy, security, and responsible AI governance. Intuit safeguards customer data and protects privacy using industry-leading technology and practices. We adhere to responsible AI principles that guide how the company operates and scales its AI-driven expert platform with its customers' best interests in mind. Our multi-disciplinary approach to governance includes internal review, executive-level oversight through our AI Governance Committee, employee training, and channels for stakeholder feedback
As AI becomes increasingly intertwined with our daily lives, Intuit recognizes its potential to cause unintended or harmful impact. Our internal governance ensures we thoughtfully address these impacts in the AI-driven experiences we create. We also work with third-party organizations across industry, policy, and academia to help address these difficult issues.
Since financial data is among the most sensitive, GenOS was designed with Responsible AI embedded as a fundamental principle, not just a checkbox. For example, GenSRF (security, risk and fraud) includes built-in guardrails against threats like prompt injection and data leakage, while enhancing content safety. Intuit technologists are working to expand GenSRF to include agentic workflows and create new controls to address the specific risks associated with AI agents
We've also built a robust Evaluation Service, an automated tool for evaluating LLM and AI applications using a wide range of metrics to help product teams experiment, develop and assess user experience. These evaluation services are integrated into existing workflows, empowering developers to build delightful customer experiences. Enhancements to the Agent Starter Kit with GenOS Evaluation Service include frameworks for monitoring and measuring agent performance
Q: Speed vs. safety is a classic trade-off. How did you solve it?
Jayanth: We don't think it has to be a trade-off. With GenOS, speed and safety go hand in hand.
GenOS provides a standardized, pre-approved "paved path" for development, which means security, privacy, and compliance guardrails (through frameworks like GenSRF) are built in from the start. Developers don't need to stop and figure out how to implement safety each time. The heavy lifting has already been handled at the platform level.
Tools like the Agent Starter Kit and the GenOS AI Workbench provide the foundation, equipping teams to build securely without sacrificing speed. That frees them up to focus 100% on solving customer problems and delivering innovation. That’s how we move quickly while maintaining trust—and in financial software, trust is absolutely essential.
Moneycontrol Journalists are not involved in creation of this article.
Discover the latest Business News, Sensex, and Nifty updates. Obtain Personal Finance insights, tax queries, and expert opinions on Moneycontrol or download the Moneycontrol App to stay updated!
