At first, robotic process automation coupled with low-code platforms and orchestration tools propelled many organizations to increase productivity and scale business operations. Virtual agents and chatbots then took automation one step further by enabling a conversational experience. Then, large language models (LLMs), vector databases, retrieval augmented generation (RAG), and other generative AI innovations enabled new ways to summarize content, generate code using copilots, and answer questions conversationally.
AI agents combine automation, conversational experiences, and process orchestration capabilities to lead us to the next phase of generative AI evolution and digital transformation. They provide developers, business users, and others with a role-based partner, proactively automating steps and acting as knowledgeable collaborators in getting work done. Integrating genAI technologies with role-based workflows is a key opportunity to deliver transformational generative AI business benefits beyond productivity improvements.
“AI agents are fundamental to practical, measurable applications of generative AI in the enterprise,” says Simon Margolis, Associate CTO of AI/ML at SADA. “Their unique ability to act—that is, write data and make API calls—represents a huge advantage in how businesses can not just gain information from their AI tools but use those tools to perform actions that are otherwise poor uses of human time.”
Platforms such as Appian, Atlassian, Cisco Webex, Cloudera, Pega, Salesforce, SAP, ServiceNow, and Workday announced AI agent capabilities this year, while public cloud agents such as Amazon Q Developer are embedded in the developer experience. The AI agent market size was valued at $3.86 billion in 2023 and is expected to grow at a compound annual growth rate of 45.1% from 2024 to 2030.
One reason for the massive interest in AI agents is that they bring expertise and automation to the workflows end-users perform regularly.
Margolis adds, “Action agents are pivotal in helping organizations realize the measurable benefits of generative AI, whether it’s a simple sales agent asking a salesperson basic customer information to automatically make entries into a CRM system or a medical agent providing information about a patient and updating their records post-visit.”
Rules-based chatbots vs. AI agents
Virtual agents and chatbots are often rule-based approaches to help end-users solve a handful of basic problems. For example, IT services management (ITSM) chatbots often address common service requests such as password resets and unlocking accounts but then redirect users to FAQs and knowledge bases for more complex requests. ITSM AI agents can perform more sophisticated tasks like predictive incident management, intelligent ticket routing, and problem root-cause analysis.
“AI agents are changing the game across industries by automating tasks, solving problems, and improving workflows,” says Abhi Maheshwari, CEO of Aisera. “Unlike standard chatbots, these agents can reason, plan, and take action independently. They’re used in areas like tech, manufacturing, legal, retail, education, and government.”
Many platforms now have sidebars on webpages and other user experience elements where end-users can interact with AI agents around their work. Sometimes, the agent presents information proactively so people can take action. At other times, they lend expertise and share data-driven insights with the employee while performing their work.
“To the user, the app interface of chatting with a bot feels familiar, as if they are chatting with one of their colleagues or in a group chat,” says Gaurav Kumar, VP at LatentView Analytics. “The difference is that the chatbots’ responses are succinct, insightful, instantaneous, and derived from a vast corpus. An additional benefit of this powerful technology is the ability to control who gets to see what using role-based access controls.”
Parul Mishra, VP of product management in digital labor at IBM, says enterprise AI agents and assistants will coexist along a continuum, with AI assistants executing prescriptive tasks while the more self-directed AI agents reason through complex problems and execute multi-step plans to solve them. Mishra says, “While the capabilities of AI agents across specific domains will vary depending on industries and use cases, the key point is that those agents will work in collaboration with AI assistants alongside a suite of other tools to transform a chatbot experience into a multi-dimensional system that can plan, test, write, and autonomously implement solutions.”
AI agents require high-quality data
To be helpful AI agents need accurate, relevant, and up-to-date information to provide accurate answers. Before leveraging AI agents, data leaders should learn what data the AI agent accesses and validate its quality.
“AI agents are autonomous systems and workflows that help automate complex tasks and decision-making; however, their efficacy relies on having access to high-fidelity data as input,” says Abhas Ricky, chief strategy officer at Cloudera. “If agents take action on unverified data, it can inadvertently introduce errors or inefficiencies into critical operations, ranging from fraud detection to supply-chain management. Ensuring trusted data integrity and accessibility through robust data architectures allow agents to make accurate, impactful decisions.”
Organizations should update their AI governance and data governance policies to include AI agent use cases. Now is also the time to review whether data pipelines need performance or other operational improvements. Larger enterprises looking to develop AI agents should review data fabrics to simplify access to datasets across SaaS, public cloud, and data centers.
“We’ll also see an increase in the need for roles focused on AI governance, helping to ensure that virtual agents don’t go rogue with unintended bias, which could result in serious implications for brands,” says Don Schuerman, CTO of Pega. “Ultimately, the future of work with virtual agents provides more opportunities for enterprises to quickly help their customers while enabling their employees to be more productive with higher-value work, all while operating at increased levels of efficiency and effectiveness.”
AI agents can improve the employee experience
AI agents offer productivity benefits, but even more importantly, they can add joy to an end-user’s workday. Instead of being bogged down by all the technical minutia needed to complete a task, employees can focus on what needs to be done while the AI agent collaborates on the implementation.
“AI agents will not only simplify human resource and finance processes, but they will transform how work gets done,” says David Somers, chief product officer of Workday. “By automating routine tasks such as submitting expenses, creating job descriptions, and scheduling, AI agents will allow employees to work more efficiently and focus on higher-value activities. This will help organizations save time and make more informed decisions without adding complexity to their workflows.”
Some quicker wins for deploying AI agents involve taking the drudgery out of work, especially for infrequently performed tasks requiring data entry.
Using HR and finance AI agents can improve everyone’s productivity because all employees submit forms and reports to these functions. Another quick win is in customer success and support, where providing accurate and timely answers can be challenging when there’s a lot of information to review.
“A key use case is semi-autonomous customer success agents, which augment and bring value to customer success teams by automating tasks instead of just advising or assisting on them,” says Raju Malhotra, chief product and technology officer at Certinia. “AI agents also play a big role for customer support, resource managers, and project managers by amplifying the impact they would not be able to have otherwise.”
Technologists who want to understand the value of AI agents should review how they work in ITSM and devops practices. While copilots can help generate code, AI agents aim to perform tasks and reduce complexity across the full software development and deployment lifecycles (SDLC).
“Virtual Agents are already transforming devops practices across all industries by providing productivity gains across all five phases of the SDLC,” says David Brooks, EVP of evangelism at Copado. “We are in a transition period from copilots to true agents, where these virtual team members can not only provide advice but perform actions on your behalf. Agents will help free up people to focus on prototyping with the help of a plan and build agent until results meet expectations, and then devops agents can test, release, and operate the solution.”
AI agents for industry-specific use cases
Many currently available AI agents target departmental workflows and aim to improve productivity. AI agents aiming to solve industry-specific workflow challenges may prove to be of even greater business value as they can directly impact costs, quality, scalability, and customer experience. These AI agents tap into the enterprise’s data and subject matter expertise to solve real employee and customer challenges while improving experiences.
“Retail, manufacturing, and any industry with large product inventories will benefit from virtual agents that combine AI, an understanding of the buyer’s needs, and deep context awareness for thousands if not millions of products,” says Jonathan Taylor, CTO of Zoovu. “This solves two of the biggest issues in online buying today—choice overload and the high cost of returns due to low-quality product search and discovery.”
AI agents that solve customer pain points and pinpoint relevant information from big data sources can provide business value, especially when AI agents are personalized to end-user roles and interests.
“In retail, virtual agents that understand a shopper’s needs will interact with retailers’ websites using machine-to-machine protocols to buy products based on the customer’s interests, price, size, and other factors,” adds Taylor.
The growing list of opportunities to use AI agents to accelerate industry 5.0 target work safety, predictive maintenance, and improving quality. Advancing in these and other areas will require IT to accelerate efforts to centralize and cleanse operational data.
“Manufacturing and physical industries will lag behind knowledge industries in developing and adopting AI-powered agents due to the lack and complexity of data infrastructure,” says Artem Kroupenev, VP of Strategy at Augury. “That being said, we’re starting to see industrial AI agents take shape around machine health diagnostics and production health recommendations because these types of solutions are narrowly focused on specific high-value use cases and come as a full-stack package with built-in hardware and data infrastructure required to make these AI agents effective.”
AI agents in manufacturing will add to existing machine learning capabilities and provide workers with two-way human-to-machine interfaces using natural language prompting.
“With AI agents for machine and production health, we’re already seeing a preview of what’s possible for AI in manufacturing: dramatic impact on productivity, increased capacity and sustainability, increased safety and augmentation of human decision-making and collaboration on the production floor,” adds Kroupenev.
Kevin Miller, CTO of Americas at IFS, shares a common problem in manufacturing that frustrates workers and impacts operations and costs. “Factory workers and field technicians often spend significant time troubleshooting unique problems not documented in any manual,” says Miller. “AI-enabled customer support tools enable predictive and prescriptive maintenance to minimize unexpected machine downtime or service interruptions.”
Risks and opportunities using AI agents
AI agents are relatively new, which brings both opportunities and risks for early adopters.
Eilon Reshef, co-founder and chief product officer at Gong, says AI agents are currently unsuitable for more comprehensive tasks that require extensive knowledge and history. Reshef says, “Asking an agent to share key takeaways from a call is not the same as asking it to plan next quarter’s revenue strategy, and the potential for unpredictable outcomes grows with the amount of autonomy you give the agent. Today’s ‘AI agents’ can be considered revolutionary for internal process automation, but enterprise leaders may need to think twice about having them take action that can be seen or felt outside of the organization.”
If AI agent capabilities improve at the same velocity as LLMs, we will see them become more autonomous and capable of agent-to-agent integrations.
Maheshwari of Aisera says, “The next big leap will be autonomous AI agents that can manage entire processes without human intervention, leading to major productivity boosts. These agents will handle routine tasks, allowing organizations to work smarter and faster. While challenges remain, this technology is set to transform how we work.”
“Eventually, AI agents will interact with each other behind the scenes, automating many human processes as one “agentic” system,” adds Deon Nicholas, CEO of Forethought. We’ll likely see a network of AI agents, similar to how the Internet today is a network of computers. Each person will have an AI agent that can book haircuts or order groceries, and those shops will have customer-experience AI agents to receive the orders, process returns, and take actions.”
Platforms are already delivering autonomous and AI agent integrations. One example is the recently announced Workday and Salesforce integration between their AI agents.
Our mainstream view of workflow focuses on task automation, process orchestration, machine learning intelligence, and human-in-the-loop decision-making. Adding AI agents to this mix will usher in a new era of digital transformation by reengineering business processes and customer experiences.