Artificial intelligence (AI) agents represent the next frontier of AI’s rapid evolution, amplifying its benefits. Tech leaders and commentators have described 2025 as the “year of AI agents,” and it’s clear the next frontier is well underway.
While AI agents have garnered public attention for their B2C applications, such as the ability to plan vacations to popular beach destinations, their impact on business productivity and the economy are less well-known. The White House AI Action Plan — released last week — focused on “winning the race” in AI, and enabling AI agent adoption is a critical part of how America can win.
AI’s global economic impact is surging, with investments expected to reach over $22 trillion by 2030 and AI adoption fueling up to a 15 percent increase in GDP by 2035. With 78 percent of companies indicating they plan to implement AI agents, AI’s benefits and economic impact will likely only continue to increase. As enterprise software companies increasingly use AI agents to enhance their products and services, they’re not only optimizing various business functions; they’re catalyzing the next phase of digital business transformation.
A look behind an AI agent’s capabilities and operation illustrates why these advances are cutting-edge. AI agents are software systems powered by large language models (LLMs) – the technology behind popular generative AI tools that can respond to questions or generate other content. But AI agents don’t simply predict an outcome or generate content; they reason and take action, providing limitless opportunities. AI agents can:
- Sense their environment through sensors or APIs to access relevant information from external systems, databases, and web services;
- Reason, often using machine learning to understand data and context;
- Plan how to accomplish an objective;
- Coordinate with users or other systems;
- Act; and
- Adapt.
The Business Software Alliance’s (BSA) new “Enterprise AI Agent” paper highlights key industry use cases for AI agents, including customer relationship management and core business processes, like human resources, finance, IT, content management, manufacturing, cybersecurity, and software development. Notable examples include:
- Customer Relationship Management: Elevate priority tickets with certain criteria for resolution, addressing the most important issues more quickly;
- Business Processes and Workflow: Provide employees with intelligent, personalized growth opportunities to increase internal mobility; expedite contract management by analyzing agreements, flagging issues, and surfacing areas that require human expertise; provide customized workflows for specific project tracking and strategic planning needs; or schedule and summarize meetings on a video communications platform;
- Manufacturing: Increase the efficiency of product design, streamline manufacturing operations, and automate error resolution; and
- Cybersecurity: Reduce incident response time by automating security threat detection, investigation, and response.
Whether it is an IT agent identifying a network problem, proposing a solution, and fixing it after human approval, or an industrial AI agent assisting with manufacturing design, enterprise AI agents are transforming the competitive landscape for global businesses.
AI Agents: Key Policy Considerations in Focus
As companies increasingly develop and use AI agents, an overarching principle in responsible AI governance remains important: roles and responsibilities matter. Different AI actors perform different tasks, which affects the information they have access to and their ability to resolve an issue. As a result, obligations for responsible AI development and use should fit the role, which is something BSA has emphasized for several years.
The AI agent ecosystem is complex, varied, and evolving. One scenario includes LLMs developers, companies integrating LLMs into their own AI agentic systems (“integrators”), and deployers, who use the agents in their business, such as a bank using an agent to enhance its fraud detection or a retailer building an agent on an integrator’s platform to assist with customer interactions. As BSA highlighted in its blog, “Unpacking the AI Stack,” role-based responsibilities are critical for ensuring appropriate steps for AI actors are feasible to implement in practice. For example, integrators, who incorporate AI tools from developers into software applications for other businesses, should have tailored responsibilities that are proportionate to the changes they make or the risks they introduce.
BSA members also continue to prioritize key safeguards in agentic AI, such as privacy and security, ensuring AI agents act in compliance with data policies and appropriate permissions and access controls. Notably, cybersecurity providers not only maintain existing network security operations, they also secure AI agents to, among other things, protect against malicious inputs and sensitive data leaks.
In short, leading enterprise software companies are propelling AI-enabled digital business transformation with AI agents, all while prioritizing privacy and cybersecurity. As stakeholders assess responsibilities within the AI agent ecosystem, an enduring principle can help guide AI development and use: responsibilities for AI actors, including integrators, should fit their role. This role-based approach ensures that appropriate safeguards can be put in place by the right people at the right time, building trust throughout the ecosystem and spurring continued AI adoption. As we continue to accelerate in the AI race, enterprise AI agents are key to crossing the finish line.