Yours For the past several years, it has been on a trip with generous AI, which includes technology as part of its services in Quickbooks, Credit Karma, Turbotax and Mailchimp.

Today the company is taking the next steps with a series of AI agents, which is to move beyond how to change small and middle-market businesses. These new agents act as a virtual team that automatically automatically makes the workflows and provides a real -time professional insight. They include payment, accounts and capabilities for finance that will directly affect business operations. According to Intuit, customers save up to 12 hours per month and on average, new agents will be paid rapidly up to five days.

“If you look at the trajectory of our AI experiences in Intuit in the early years, AI was made in the background, and with Intuit aidYou saw a change to provide information back to the customer, “Chief Data Officer Ashok Srivastava told Venturebeat at Intuit.” Now what you are seeing is a complete redesign. Agents are actually working on behalf of the customer, with their permission. ,

Technical architecture: from starter kit to production agents

Intuit has been working on the way from assistant to agent AI for some time.

In September 2024, company Discretion your plans To use AI to automate complex functions. It is a firm approach on the company’s generic AI operating system (Genos) platform, the foundation of its AI efforts.

Earlier this month, Intuit announced several attempts that further their abilities. The company has developed its own Early adaptation service This will optimize questions for any major language model (LLM). It has also developed what it says to an intelligent data cognition layer for enterprise data that can understand the various data sources required for enterprise workflows.

Moving one step ahead, Intuit developed an agent starter kit, which creates the company’s technical foundation to enable agent AI development.

Agent portfolio: from cash flow to customer management

With the technical foundation, including the agent starter kit, intuit has created a series of new agents that help business owners achieve things.

The agent of the intuit displays the technical sophistication required to go for autonomous workflow execution from AI, a future. Each agent coordinates the prediction, natural language processing (NLP) and autonomous decision making within full commercial processes. they include:

Payment agent: Autonomously optimize the cash flow by predicting late payment, generating invoices and performing follow -up sequences.

Accounting agent: Rules represents the development of intuit from system-based systems to autonomous bookkeeping method. The agent now automatically distributes transaction classification, reconciliation and workflow, cleaner and more accurate books.

Finance agent: Traditionally dedicated business intelligence (BI) automates the strategic analysis for the need of equipment and human analysts. The major performance indicator (KPI) analysis, landscape plan and forecast that the company is autonomy how to generate development recommendations against the benchmark.

Intuit is also creating customer hub agents that will help in customer acquisition works. Along with payroll processing, project management efforts are also part of future release plans.

Beyond interactive UI: work-oriented agent design

New agents mark a development as to how AI users are presented.

The interface of the intuit has revealed the important user experience principles for entering the enterprise agent to purpose. Instead of speaking AI capabilities on existing software, the company fundamentally reorganized the quickbook user experience for AI.

“The user interface is now really oriented around business functions, which need to be done,” Srivastava explained. “This allows real -time insight and recommendations to come directly to the user.”

This working-focused approach dominates the current venture AI tool with a chat-based interface. Instead of users’ needs to make strategies or navigate the convenient flow to users, agents work within the existing business workflows. The system involves the intuit a “business feed” that relevantly gives the surface to agent actions and recommendations.

Faith and Verification: Closed Loop Challenge

One of the most technically important aspects of the implementation of intuit addresses a significant challenge in the autonomous agent regulation: verification and trust. Enterprise AI teams often struggle with the black box problem – how do you ensure that AI agents are performing correctly when they work autonomously?

Srivastava insisted, “To create a trust with Artificial Intelligence System, we need to provide proof points to the customer that what they think is actually happening.” “That closed loop is very, very important.”

The solution of intuit involves the construction of verification capabilities directly in genos, allowing the system to provide evidence of agent functions and results. For the payment agent, this means that users showed that the challans were sent, delivery was tracked and the payment cycles that gave results from the functions of the agent were demonstrated.

This verification approach provides a template for enterprise teams that deploy autonomous agents in high-day commercial processes. Instead of asking users to rely on the AI ​​output, the system provides the result of the auditable trails and average.

What does it mean for enterprises to go to Agent AI

The development of intuit provides a solid roadmap for the enterprise teams planning autonomous AI:

Focus on completion of workflow, no conversation: Target specific business processes for end-to-end automation rather than the construction of general-purpose chat interfaces.

Create agent orchestration infrastructure: Invest in platforms that coordinate prediction, language processing and autonomous execution within integrated workflows, not isolated AI tools.

Design verification system upfront: Include comprehensive audit trails, utkam tracking and user information as core capabilities rather than core capabilities.

Map workflows before construction technology: Use customer advisory programs to define agent capabilities based on real operating challenges.

Plan for Interface Redigation: Adapt the UX for agent-operated workflow instead of traditional software navigation pattern.

Srivastava said, “As big language models decrease, the experiences that are created on them become very important.”