Leave AI ‘Bake-off’ and create autonomous agents: Lessons from Intuit and Amex

Due to the generative AI mature, enterprise is transferred to the implementation by use – intelligent, autonomous agents beyond chatbots and copylots. In a conversation Ventricular Matt Marshal, Ashok Srivastava, SVP and Chief Data Officer YoursAnd Hillary Packer, EVP and CTO American Express But VB conversionIt is wide how their companies are embracing agents AI to change customer experiences, internal workflows and core business operations.

Model to Mission: Emergence of intelligent agents

In intuit, agents are not only about answering questions – they are about executing tasks. For example, in turbotax, agents help customers to complete their taxes 12% faster, about half the finish in less than an hour. These intelligent systems draw data from several streams-including real-time and batch data-internal bus and frequent services of intuis. Once processed, the agent analyzes the information for decision making and action.

“This is the way we are thinking of agents in the financial domain,” Srivastava said. “We are trying to ensure that we are constructed, they are strong, scalable and really anchor in reality. The agents we are experiencing are designed to work. For Customer, with Their permission. It is important to create faith. ,

These capabilities have been made possible by the custom generative AI operating system of genos, intuis. It is clear in its heart, which Srivastava prefers from a CPU: it receives data, causes causes on this, and determines an action that is then executed to the final user. The OS was designed to remove technical complexity, so developers do not need to reinforce the risk safety measures or safety layers every time they build an agent.

Among the brands of Intuit – from Turbotax and Quickbooks to Mailchimp and Credit Karma – Genos helps create reliable, reliable experiences and ensures strength, scalability and excellence in cases of use.

Construction of Agent Stack in Amex: Trust, Control and Experimentation

For Packer and his team in Amex, the steps in agent AI creates more than 15 years of experience with traditional AI and a mature, war-tested large data infrastructure. As genai capabilities accelerate, Amex is re -shaping its strategy to focus on how intelligent agents can run internal workflows and strengthen the next generation customer experiences. For example, the company focuses on developing internal agents that promote employee productivity, such as the APR agent who reviews the software bridge requests and advises engineers whether the code is ready to merge. The project reflects the comprehensive approach of Amex: start with matters of internal use, proceed quickly, and use the initial win to refine the underlying infrastructure, equipment and governance standards.

To support rapid use, strong security and policy enforcement, Amax developed a “competent layer”, which allows rapid growth without renunciation of the overs. “And so now as we think about the agent, we have found a good control aircraft to plug into these extra, additional railings that we really need to keep in place,” Packer said.

Within this system there is the concept of the modular “brain” of the amex – an outline in which agents are required to consult a specific “brain” before taking action. These brain serve as modular governance layers-Brand values, privacy, security and legal compliance-that each agent must be attached to the decision making. Each brain represents a domain-specific set of policies, such as brand voice, privacy rules, or act as legal obstacles and acts as an advisory authority. By rooting decisions through this system of obstacles, agents remain accountable, combine with enterprise standards and users are worthy of trust.

For example, a dining reservation agent, who is working through the restaurant booking platform of Reji, Amex, must be valid that it is selecting the right restaurant at the right time, following the brand and policy guidelines, matches the user’s intentions.

Establishment that enables speed and security

Both AI leaders agreed that enabling rapid growth on scale leads to a demand for thoughtful architectural design. In intuit, Genos builds hundreds of developers the right to build safe and continuously. The platform ensures that each team can work without using shared infrastructure, general safety measures and models flexibility.

Amax took a similar approach with its competence layer. Designed around an integrated control aircraft, layer teams apply centralized policies and railings to develop AI-operated agents faster. This ensures the risk and frequent implementation of the governance structure, encouraging speed. Developers can quickly deploy experiments, then without compromising with the brand Trust, evaluating and scale on the basis of reaction and performance.

Agent AI adoption lesson

Both AI leaders emphasized the need to move forward quickly, but with intentions. “Don’t wait for a bake-off,” Packer advised. “It is better to choose a direction, obtaining something in production, and rather than repetitively, instead of delaying the correct solution that may be older than the launch time.” He also emphasized that the measurement should be embedded from the beginning. According to Srivastava, the instrumentation is nothing to bolt later – it should be an integral part of the stack. Tracking cost, delay, accuracy and user impact are necessary to assess the price and maintain accountability on the scale.

Srivastava said, “You must be able to measure it. This is the place where genos arrives-one has an inherent capacity that allows us to track the instrument AI application and both and track the costs in the return.” “I review every quarter with my CFO. We go to line from the line through every AI use case in the company, assess how we are spending and what value we are getting in return.”

Intelligent agents are the next enterprise platform shift

Intuit and American Express are one of the leading enterprises, who are only adopting agents AI as a technology layer, but as a new operating model. Their approach focuses on the construction of the agent forum, the establishment of the governance, measuring the impact and proceeding quickly. As the hopes of the enterprise develop from simple chatbot functionality to autonomous execution, the agent will be the best place to lead AI as a first-class discipline as a first-class discipline, as well as the best place to lead the race of agent along with the agent.

Editor’s Note: As a thanks to our readers, we have opened early bird registration for VB Transform 2026-$ 200. This is the place where AI is the ambition operational reality, and you want to stay in the room. Now reserve your place,

Daily insights on business use cases with VB daily

If you want to impress your boss, VB daily has covered you. We give you the scoop inside what companies are doing with generative AI, from regulatory changes to practical deployment, so you can share insight for maximum ROI.

Read our privacy policy

Thanks for membership. See more VB newsletters here.

There was an error.