Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Join the event that trusts business leaders for almost two decades. VB Transform brings together people who build a real business AI strategy. Learn more
Intuity has made a trip in recent years with a generative AI, incorporating technology as part of its services at QuickBooks, Credit Karma, Turbotax and Mailchimp.
Today, the company is taking the next step with a series of AI agents that go beyond to transform the operation of small and medium-sized enterprises. These new agents work as a virtual team that automates workflows and provides real -time commercial information. They include payment, account and finance capacities that will have a direct impact on commercial operations. According to Intuit, customers save up to 12 hours per month and, on average, will be paid up to five days faster thanks to new agents.
“If you look at the trajectory of our IA experiences in intuits in the first years, the AI was integrated into the background, and with Intuit assistYou have seen a change to provide information to the customer, “said Ashok Srivastava, AI and the data agent of Intuit.” Now what you see is a complete overhaul. The agents actually work on behalf of the customer, with their permission. »»
Technical architecture: from starter kit to production agents
Intuits has been working on the path of agenic AI assistants for some time.
In September 2024, the company detailed his plans To use AI to automate complex tasks. This is a solid approach on the generative platform of the company’s operating system (Genos) of the company, the basis of its AI efforts.
Earlier this month, Intuits announced a series of efforts that still extend its capacities. The company has developed its own Quick optimization service This will optimize requests for any large language model (LLM). He has also developed what he calls an intelligent data cognition layer for corporate data that may include different data sources required for business workflows.
Going a little further, Intuits has developed an agent starter kit that is based on the technical foundations of the company to allow the development of agentic AI.
The portfolio of agents: cash flows to customer management
With the technical foundation in place, including agent start -up kits, Intuit has built a series of new agents that help business owners get things done.
The continuation agent follows the technical sophistication required to move from predictive AI to the autonomous execution of the workflow. Each agent coordinates prediction, natural language treatment (NLP) and autonomous decision -making in complete business processes. They include:
Payment agent: Autonomously optimizes cash flows by predicting late payments, generating invoices and performing follow -up sequences.
Accounting agent: Represents the evolution of intuits of systems based on rules for autonomous accounting. The agent now manages the categorization of transactions, reconciliation and the completion of workflow, the delivery of cleaner and more precise books.
Financial agent: Automation strategic analysis traditionally requiring commercial intelligence tools (BI) and human analysts. Provides an analysis of key performance indicators (KPI), scenarios planning and forecasts depending on the way the company is against peers while autonomously generating growth recommendations.
Intuits also builds Customer Hub Agents who will help customer acquisition tasks. Payroll treatment as well as project management efforts are also part of future publication plans.
Beyond the conversational user interface: agent design focused on tasks
The new agents mark an evolution in the way AI is presented to users.
The redesign of the induced interface reveals important user experience principles for the deployment of business agents. Rather than bolling AI’s capacities on existing software, the company has fundamentally restructured the QuickBooks user experience for AI.
“The user interface is now really oriented around the commercial tasks that must be carried out,” said Srivastava. “It allows real -time information and recommendations to come directly to the user.”
This approach centered on the task contrasts with the interfaces based on cat dominating the current tools of corporate AI. Instead of asking users to learn incitement strategies or navigate conversation flows, agents operate in existing commercial work flows. The system includes what Autuits calls a “commercial flow” that faces the actions and recommendations of agents contextually.
Confidence and verification: the challenge in a closed loop
One of the most technically significant aspects of the implementation of intuits takes up an essential challenge in the autonomous deployment of agents: verification and confidence. Corporate AI teams often have trouble with the black box problem – how do you make sure that AI agents work properly when they work independently?
“In order to strengthen confidence with artificial intelligence systems, we must provide proof points to the customer that what they think of happens really happens,” said Srivastava. “This closed loop is very, very important.”
The intuit solution implies strengthening the verification capacities directly in the genos, allowing the system to provide evidence of the actions and results of the agents. For the payment agent, this means showing users that invoices have been sent, monitoring delivery and demonstration of the improvement of payment cycles resulting from the agent’s shares.
This verification approach offers a model for company teams deployment of autonomous agents in business processes with high issues. Rather than asking users to trust AI outings, the system provides verifiable trails and measurable results.
What this means for companies that seek to enter the AI agent
The evolution of Intuit offers a concrete roadmap for business teams planning autonomous IA implementations:
Focus on the completion of the workflow, not the conversation: Target specific business processes for end -to -end automation rather than creating cat interfaces for general use.
Build an agent orchestration infrastructure: Invest in platforms that coordinate prediction, language treatment and autonomous execution in unified work flows, not isolated AI tools.
Design verification systems in advance: Include complete audit trails, results monitoring and user notifications as a basic capacities rather than therefore.
Workflows card before construction technology: Use customer advice programs to define the capacities of agents according to real operational challenges.
Plan the overhaul of the interface: Optimize UX for workflows focused on agents rather than for traditional software navigation models.
“As important language models become merchants, the experiences built there are much more important,” said Srivastava.
Source link
