10 ways Intuit is revolutionizing personalization with generative AI

Intuit's generative AI strategy reflects the company's vision to help consumers and small businesses be more competitive and confident in every financial decision. Core to that vision is a heavy reliance on generative AI to personalize customers' experiences with Intuit products, automate routine tasks, and provide new insights to help customers achieve their financial goals. The company has more than 100 million customers worldwide through its TurboTax, Mint, Credit Karma, QuickBooks, and Mailchimp products.

Intuit's AI-driven platform strategy is working, as evidenced by double-digit growth in both total revenue and operating profit. Total revenue grew to $6.7 billion in the third quarter of fiscal 2024, which ended April 30, up 12% from the prior year. Also, revenue from the company's small business and self-employed group surged 18% to $2.4 billion.

How Intuit is revolutionizing personalization with GenAI

In a fascinating interview at VB Transform 2024, Nhung Ho, VP of AI at Intuit, spoke with Matt Marshall, CEO and Editor-in-Chief of VentureBeat, about how Intuit continues to refine its approach of using gen AI, AI, and machine learning (ML) modeling to fine-tune its personalization strategies by customer segment, product, and development strategy.

Ho provided insight into the practical challenges along the way and talked about how AI-based automation is helping Intuit achieve its twin goals of simplifying personal finances while empowering small business customers to be more efficient and informed.

Intuit is committed to democratizing AI to the fullest extent possible across all of its applications, empowering small businesses with real-time data and better insights to find new revenue opportunities while retaining much of their bottom line. In September 2023, the company released Intuit Assist, its first-generation AI-powered financial assistant for small businesses and consumers. The tool integrates with and leverages Intuit's proprietary Generative AI Operating System (GenOS). It is designed to provide personalized financial insights and recommendations across products such as TurboTax, QuickBooks, Credit Karma, and Mailchimp.

In a wide-ranging interview with VentureBeat last year, Intuit's Chief Data Officer Ashok Srivastava likened the GenOS infrastructure to an operating system for gen AI. “Think of an actual operating system like MacOS or Windows,” he said of the assistant, management, and monitoring capabilities. Similarly, large language models (LLMs) need a way to coordinate their actions and access the resources they need. “I think this is a revolutionary idea,” Srivastava said.

Gen AI turbocharges TurboTax

“We've spent a lot of time over the past year experimenting with incorporating AI applications into a variety of products, but the one I want to talk about is TurboTax, which has been deployed at the largest scale. Last tax season, 24 million customers had a generative AI experience that helped them fully understand how their taxes were calculated and how they got there,” Ho told the Transform audience.

“So we all file our taxes hopefully before the end of the year, but a lot of people, myself included, when the end of the year rolls around, are left wondering, did I do it right?,” she continued. “How did I get here? What got me to this point? And in the past, there just wasn't a good way to interpret and make sense of all those massive numbers and calculations in a way that a regular person like me could understand.”

Ho explained why Intuit relies on Anthropik's LLM to provide accuracy. “What we did was combine the power of our tax engine, which ensures accuracy, with an LLM, specifically Claude from Anthropik, to help explain how to do the calculations,” she said. “And I think that was the magic that we were able to provide to our customers last season. I think it was a really great moment of confidence for them to say, 'I know what's going on. I don't need to be scared. I don't need to worry that I've done something wrong. I feel like I'm part of this process and I'm kind of taking my power back in the scanning process.'”

Intuit has fine-tuned its approach to using Generation AI to provide a higher level of personalization based on unique customer requirements. Ho explained to the Transform audience: “To give an example in the small business space, personalization means using your data to modify our product and tailor it to your situation. A construction company and a bakery have very different ways of setting up their business and they also have a very different set of services that they use us for. When a customer logs in and signs up for our product and tells us what they're up to, we start personalizing them right away and making sure we only use the parts that are most relevant to them, rather than everything else.”

Based on the insights shared during the discussion, it's clear that Intuit relies on a set of core actions to make personalization great across all of its products using gen AI. These include:

Integrating AI to simplify complex tasksIntuit is using gen AI to break down complex tasks into steps that users can understand, as evidenced by the AI-driven explanations in TurboTax that simplify tax calculations for its 24 million customers.

Automate routine tasks: Intuit wants to integrate gen AI into any app to automate mundane tasks, freeing up its customers' employees to spend more time on their core responsibilities, including talking to customers. A prime example is how its AI tools can automate accounting and marketing workflows for small businesses.

Ensuring high-quality data: According to Ho, maintaining high-quality, clean data is essential for AI applications to work effectively. Intuit continuously monitors and curates data to ensure accuracy in its AI-driven processes. “High-quality data is what gives you a differentiating advantage. Do you have the right data? Is it clean? Is it actually correct?” asks Ho.

Managing Latency in AI Systems: Overcoming latency bottlenecks in AI apps is essential to successfully delivering user experiences at scale. Intuit is researching, and currently using, AI apps outside of its chat-based interface to reduce its sensitivity to latency. “These systems are slow. We live in an age of low patience, and we're used to things working in milliseconds. In the world of generative AI, you're talking tens of seconds,” Ho noted.

Embracing a Multi-Cloud Strategy: Adopting a multi-cloud approach has allowed Intuit to maintain flexibility and access the best AI models for specific personalization use cases. This strategy also allows them to leverage the best tools available, regardless of cloud provider. “We chose to be completely agnostic, it's just APIs, so it doesn't matter. Whichever cloud provider builds the best ecosystem for this, that's a win for me,” Ho advised.

Building Scalable and Secure AI SystemsIntuit is focusing on data scalability as a core part of its AI security strategy. Hardening AI systems against attacks also lays the foundation for more efficient innovation cycles. “We want to empower everyone to innovate, but we need to do it securely. To build a system that guarantees that, we need to ensure that our data is secure,” says Ho.

Create a composable architecture: Designing a system with fully composable microservices gives you flexibility. Intuit uses a composable architecture to integrate different AI models and technologies. “We built a fully composable set of microservices and capabilities, which means you can choose which pieces you combine to create new applications,” Ho explains.

Personalize the user experience: It's important to tweak and personalize AI apps and tools to meet the specific needs of different user segments. Intuit customizes its AI tools for all kinds of small businesses in its customer base – bakeries and construction companies have different needs, for example, says Ho.

Balancing deterministic and stochastic systemsDetermining which AI systems should be deterministic and which can be probabilistic ensures accuracy, according to Ho. Intuit uses deterministic systems for critical tasks like calculating taxes. “We have a separate engine that does the tax calculations. It's deterministic. We'll never introduce a probabilistic system to calculate taxes,” Ho assured the audience after an attendee asked about the accuracy of Intuit's AI models in calculating taxes for customers.

Provide a roadmap to expand use cases for AI applications: Ho advised pushing the boundaries of AI applications beyond initial use cases to further enhance adjacent, related business processes. Intuit is applying AI to workflow augmentation and summarization tools to improve efficiency. “Beyond chat, there are many applications of generative AI. Think about all the parts of your workflow where you can apply AI to benefit your customers today and in their lives to actually do the work for them,” Ho recommended.

Conclusion

Intuit's forward-thinking approach to using gen AI for personalization stands out in how it uses models to further refine preferences and customer experiences by product and within specific customer segments. The release of Intuit Assist and GenOS reflects how far the company is going in using gen AI to extend and improve customer experiences.