As the landscape of artificial intelligence evolves, enterprises are transitioning from mere experimentation to practical implementation. The focus has shifted from basic chatbots to advanced autonomous agents that are revolutionizing customer experiences, internal processes, and core business functions.
At Intuit, the concept of intelligent agents goes beyond just providing answers – it’s about task execution. For example, in TurboTax, these agents assist customers in completing their taxes efficiently, resulting in a 12% faster process with nearly half of the users finishing within an hour. These sophisticated systems aggregate data from various sources through Intuit’s internal infrastructure and services. The collected data is then analyzed by the agent to make informed decisions and take appropriate actions.
Ashok Srivastava emphasizes this approach by stating,
“We’re trying to make sure that as we build, they’re robust, scalable and actually anchored in reality.”
This emphasis on building trust with customers through efficient task completion underscores the importance of these agentic experiences.
Intuit’s success in creating such capabilities can be attributed to GenOS, its custom generative AI operating system. Within GenOS lies GenRuntime which acts as the processing unit for data analysis and decision-making before executing actions for end-users. This operating system simplifies technical complexities for developers by providing a standardized framework that ensures consistency across various applications like TurboTax, QuickBooks, Mailchimp, and Credit Karma.
On the other hand, at American Express (Amex), Hillary Packer and her team are leveraging over 15 years of experience with traditional AI along with a robust big data infrastructure to venture into agentic AI territory. As Amex explores how intelligent agents can streamline internal workflows and enhance customer interactions significantly.
Packer explains their strategy as starting with internal use cases like developing an APR agent that reviews software pull requests for engineers. By focusing on quick wins internally while refining infrastructure along the way allows Amex to establish a solid foundation for scaling up intelligent agents effectively.
To ensure secure development practices without compromising oversight or speed at Amexan “enablement layer
” has been implemented.This layer facilitates rapid development by incorporating necessary guardrails seamlessly into the process ensuring compliance with policies related to security standards and privacy regulations.
“
Instrumentation isn’t something to bolt on later—it has to be an integral part of the stack,” stresses Srivastava highlighting the need for embedded measurement tools right from inception. Tracking metrics like cost efficiency,latency levels,and user impact is crucial not only for assessing value but also maintaining accountability across different AI applications within an organization.
Both leaders stress agility in adopting agentic AI technologies without waiting endlessly for perfect solutions.“Don’t wait for a bake-off,” advises Packer emphasizing the importance of quick iterations over delayed launches.Srivastava echoes this sentiment by underlining early incorporation of measurement practices into developmental processes rather than adding them as an afterthought post-launch.
The commitment showcased by companies like Intuitand American Express towards embracing agentic AI signifies more than just a technological advancement.It represents a paradigm shift towards treating intelligent agentsas essential components within enterprise operations.Establishing strong governance structures,enabling observability,and implementing modular governance frameworks will distinguish organizations preparedto leadinthe new eraof autonomous execution through agenticAI platforms.
Leave feedback about this