Building AI for financial software requires a different playbook than consumer AI Intuit’s The latest QuickBooks release provides an example.
The company has announced Intuit Intelligence, a system that orchestrates specialized AI agents on its QuickBooks platform to handle tasks including sales tax compliance and payroll processing. These new agents enhance the existing accounting and project management agents (which have also been updated) as well as a unified interface that lets users query data in QuickBooks, third-party systems, and uploaded files using natural language.
The new development follows years of investment and improvement at Intuit ZenOSthereby allowing the company to build AI capabilities that reduce Improve latency and accuracy,
But the real news isn’t what Intuit built — it’s how they built it and why their design decisions will make AI more useful. The company’s latest AI rollout represents an evolution based on hard-learned lessons about what works and what doesn’t when deploying AI in financial contexts.
What the company learned is sobering: Even when its accounting agents improved transaction classification accuracy by an average of 20 percentage points, they still received complaints about errors.
"The use cases we’re trying to solve for customers include tax and finance; If you make a mistake in this world, you lose trust with customers in buckets and we get it back only in spoons," Joe Preston, Intuit’s vice president of product and design, told VentureBeat.
Structure of belief: real data questions on productive responses
Intuit’s technology strategy focuses on fundamental design decisions. For financial queries and business intelligence, the system queries real data rather than generating responses through large language models (LLMs).
This is also extremely important: not all that data is in one place. Intuit’s technical implementation allows QuickBooks to receive data from many different sources: native Intuit data, OAuth-connected third-party systems such as Square for Payments, and user-uploaded files such as spreadsheets containing vendor pricing lists or marketing campaign data. This creates a unified data layer that AI agents can reliably query.
"We are actually questioning your actual data," Preston explained. "This is much different than if you simply copy a spreadsheet or PDF and paste it into ChatGPT."
This architectural choice means that Intuit Intelligence acts more as a system orchestration layer. It is a natural language interface for structured data operations. When a user asks about projected profitability or wants to run payroll, the system translates the natural language query into database operations against verified financial data.
This matters because Intuit’s internal research has exposed widespread shadow AI use. When surveyed, 25% of accountants who use QuickBooks admitted that they were already copying and pasting data into ChatGPT or Google Gemini for analysis.
Intuit’s approach treats AI as a query translation and orchestration mechanism, not as a content generator. This reduces the risk of hallucinations that have plagued AI deployment in financial contexts.
Explanation as a design requirement, not an afterthought
Beyond the technical architecture, Intuit has made explainability a core user experience in its AI agents. This goes beyond just providing the right answer: it means showing users the reasoning behind automated decisions.
When Intuit’s Accounting Agent classifies a transaction, it doesn’t just display the results; This shows logic. This isn’t marketing copy about explainable AI, this is actual UI that displays data points and logic.
"It’s about closing that trust cycle and making sure the customer understands why," Alastair Simpson, Intuit’s vice president of design, told VentureBeat.
This becomes especially important when you consider Intuit’s user research: While half of small businesses report AI as helpful, nearly a quarter have not used AI at all. The explanation layer serves both populations: building confidence for newcomers, while giving context to experienced users to verify accuracy.
The design also enforces human control at critical decision points. This approach extends beyond the interface. When automation reaches its limits or when users want validation, Intuit connects users directly to human experts embedded in the same workflow.
Navigating the transition from form to conversation
One of Intuit’s more interesting challenges involves managing fundamental changes to the user interface. Preston described it as having one foot in the past and one foot in the future.
"It’s not just Intuit, it’s the entire market," Preston said. "Even today we have a lot of customers filling out forms and looking at tables filled with data. We’re very invested in this and questioning the ways that we do things in our products today, where you’re basically just filling out, form after form, or table after table, because we see where the world is going, which is really a different form of interaction with these products."
This creates a product design challenge: How do you serve users who are comfortable with a traditional interface, while gradually introducing conversational and agentic capabilities?
Intuit’s approach has been to embed AI agents directly into existing workflows. This is not meant to force users to adopt entirely new interaction patterns. The payment agent appears with the invoicing workflow; The accounting agent enhances the existing reconciliation process rather than replacing it. This incremental approach lets users experience AI benefits without leaving familiar processes.
What enterprise AI builders can learn from Intuit’s approach
Intuit’s experience deploying AI in financial contexts brings to the fore several principles that apply broadly to enterprise AI initiatives.
Vaastu matters for faith: In domains where accuracy is important, consider whether you need content creation or data query translation. Intuit’s decision to treat AI as an orchestration and natural language interface layer dramatically reduces the risk of hallucinations and avoids using AI as a generative system.
Interpretability should be designed in, not bolted on: When trust is at stake, showing users why the AI made a decision is not optional. This requires well-thought-out UX design. This may constrain model choices.
Maintains user control confidence while improving accuracy: Intuit’s Accounting Agent improved classification accuracy by 20 percentage points. Nevertheless, maintaining user override capabilities was necessary for adoption.
Gradual transition from familiar interface: Don’t force users to leave the form for a conversation. First, embed AI capabilities into existing workflows. Let users experience the benefits before asking them to change behavior.
Be honest about what is reactive vs. proactive: Current AI agents primarily respond to signals and automate defined tasks. True proactive intelligence that makes strategic recommendations without prompting remains an evolving capability.
Address workforce concerns with tooling, not just messaging: If the purpose of AI is to augment rather than replace workers, provide workers with AI tools. Show them how to take advantage of technology.
For enterprises adopting AI, Intuit’s journey provides a clear direction. The winning approach prioritizes reliability over capability demonstrations. In domains where mistakes have real consequences, this means investing in sophistication in interactions or accuracy, transparency, and human oversight before taking autonomous actions.
Simpson summarized the challenge: "We didn’t want it to be a bolt-on layer. We wanted customers to live in their natural workflow, and the agents working for customers to be embedded in the workflow."
