Photo credit: venturebeat.com
Intuit, the financial software leader known for TurboTax and QuickBooks, is pioneering advancements in generative AI to better serve its small business clientele.
In a market saturated with AI claims, Intuit stands out with its agent-based AI framework that is producing measurable business advantages. The company has introduced what it refers to as “done for you” experiences that manage entire workflows independently, resulting in significant impacts on business operations.
At VB Transform 2024, Intuit elaborated on their generative AI operating system, known as GenOS, showcasing how it enhances personalization and user experience. Notably, in September 2024, they incorporated agentic AI workflows, optimizing functionality for both the organization and its users.
Recent data from Intuit indicates that QuickBooks Online users are experiencing payments on average five days sooner, and overdue invoices have a 10% higher likelihood of being settled in full. For small enterprises, these enhancements can be vital in maintaining cash flow, making them potentially transformative innovations.
The Technical Trinity: How Intuit’s Data Architecture Enables True Agentic AI
What distinguishes Intuit’s strategy from its competitors is its advanced data architecture tailored to facilitate agent-based AI functionalities.
The company has established what Ashok Srivastava, Chief Data Officer, describes as a “trinity” of data systems:
Data Lake: A foundational repository that stores all types of data.
Customer Data Cloud (CDC): A specialized layer designed for serving AI-driven experiences.
Event Bus: A streaming data platform that supports real-time data interactions.
“CDC acts as a serving layer for AI experiences, while the data lake functions as the comprehensive data repository,” Srivastava explained in an interview. “The agent interacts with data and can access a wealth of information for insights.”
Going Beyond Vector Embeddings to Power Agentic AI
Intuit’s architecture takes a different route from the common trend of hastily implementing vector databases that many organizations are adopting. While these technologies are integral for powering AI systems, Intuit emphasizes a comprehensive approach to achieve genuine semantic understanding.
“The critical factor remains ensuring logical and semantic comprehension of data,” Srivastava noted.
To foster this understanding, Intuit is developing a semantic data layer that sits atop its existing infrastructure. This layer provides depth and context to the data, enabling Intuit’s AI agents to grasp relationships among various data elements more accurately.
By creating this semantic layer, Intuit enhances the capabilities of its vector systems, allowing AI agents to make more insightful and context-aware decisions for users.
Beyond Basic Automation: How Agentic AI Completes Entire Business Processes Autonomously
Unlike many companies focusing AI on basic automation or scripted customer service interactions, Intuit aims to deliver fully agentic “done for you” applications that can perform complex, multi-step tasks with minimal human intervention required for approval.
For QuickBooks users, this agentic AI system assesses client payment behaviors and invoice statuses to create customized reminder messages. Business owners simply need to review and approve these messages before they are sent, thereby streamlining the payment process significantly.
Intuit is not just applying these principles externally; it is also building internal systems for processes like procurement and HR management.
“We have developed an internal agentic procurement process that employees can utilize for purchasing and travel arrangements,” Srivastava shared, illustrating how Intuit leverages its own AI innovations.
Designed for the Reasoning Model Era
A potential competitive edge for Intuit lies in its architecture, which was designed with foresight regarding the emergence of advanced reasoning models like DeepSeek.
“We built our generative runtime with the expectation that reasoning models would advance,” Srivastava explained. “We’re positioned ahead of the curve, having developed these capabilities with future advancements in mind.”
This proactive design allows Intuit to seamlessly integrate new reasoning abilities into their AI workflows as they become available, avoiding the need for extensive system redesigns. According to Srivastava, Intuit’s engineering teams are already harnessing these capabilities to enable agents to reason across various tools and data sources in innovative ways.
Shifting from AI Hype to Business Impact
Intuit’s approach highlights a significant emphasis on achieving real business results rather than merely showcasing technological prowess.
Intuit envisions that enhanced reasoning capabilities will lead to even richer “done for you” experiences, addressing a wider array of customer needs comprehensively. Each experience integrates numerous smaller tasks into a cohesive workflow solution.
Implications for Enterprises Adopting AI
Prioritize Outcomes Over Technology: Focus on addressing specific business challenges with clear improvement metrics instead of simply implementing AI for its own sake.
Design for Future Models: Create systems that can readily accommodate upcoming reasoning models to avoid disruptive overhauls.
Tackle Data Foundations First: Ensure your data infrastructure supports a thorough semantic understanding and the ability to connect various systems before deploying agents.
Develop Comprehensive Experiences: Move past basic automation to create holistic “done for you” workflows that provide complete solutions.
As agentic AI continues evolving, organizations that emulate Intuit’s example by concentrating on integrated solutions rather than fragmented AI functionalities may achieve similar substantial business outcomes, steering clear of mere technological hype.
Source
venturebeat.com