In January, OpenAI CEO Sam Altman said, “We believe that, in 2025, we may see the first AI agents ‘join the workforce’ and materially change the output of companies.” This bold prediction signals a transformative shift in how businesses will operate, as AI progresses from co-pilots — tools that assist humans — to autonomous systems capable of making decisions independently. These systems can automate workflows, interact with digital environments, and take action without human intervention, unlocking new efficiencies.
Nvidia CEO Jensen Huang reinforced this momentum, saying that AI agents are “likely to be a multitrillion-dollar opportunity.” With the market for AI agents expected to grow from $5.1 billion in 2024 to $47.1 billion by 2030, the race to build and deploy these systems is accelerating.
The Rise of AI Agent Frameworks
To meet the growing demand for AI agents, tech companies are developing new frameworks that enable AI agents to function autonomously across a variety of tasks. For example, Microsoft offers prebuilt agent applications and other enterprise capabilities through Copilot Studio and also offers enterprise developers the framework AutoGen to create AI agents.
Many agent frameworks help with tasks like navigating web pages and scheduling jobs, such as Anthropic’s computer use and OpenAI’s Operator, which can perform tasks like ordering groceries and filing expense reports. As companies continue to embrace AI agents, these frameworks will evolve to support increasingly sophisticated and autonomous systems.
Enterprise Use Cases for AI Agents
According to a recent study, 82% of companies plan to integrate AI agents in the next one to three years to develop automation and enhance efficiency.
One recent example is LinkedIn Hiring Assistant, which launched in October 2024. This AI agent ingests notes to turn into longer job descriptions, sources candidates, and even engages with them. This is just one of the many ways AI agents are poised to revolutionize enterprise operations, taking on tasks that have traditionally required significant human input.
Transforming Customer Service
One clear and immediate use case for AI agents is customer service. According to a study on conversational AI, 41% of companies already use AI-powered copilots for customer service and 60% have implemented them for IT help desks. In 2023, McKinsey looked at 5,000 customer service representatives using genAI and found that issue resolution increased by 14% an hour, while time spent handling issues went down 9%.
Now, many of those companies are looking to adopt AI agents. These agents could handle end-to-end customer interactions, such as resolving billing issues or processing refunds. For example, in November, ServiceNow released AI agents for customer service management to boost productivity by autonomously solving many employee and customer issues, while leaving humans in the loop for oversight and governance. The agent creates a step-by-step process for resolving an issue, and then executes on that plan with approvals from live agents where needed.
Another example is Sierra, which aims to fully automate a range of online customer interactions using AI agents. According to Sierra, this can make human agents 10-20% more efficient, while also independently handling 70% of cases. The company uses multiple AI models, including one that acts as a supervisor to ensure other AI systems are performing as expected. In January, Microsoft launched Microsoft 365 Copilot Chat, a rebranded version of its AI chat experience for businesses, enhanced with agentic capabilities and performing tasks like providing customer information before meetings and monitoring relevant events.
Accelerating Research and Data Analysis
GenAI is already enhancing research by allowing enterprise users to query both proprietary internal data and premium external documents conversationally, such as with AlphaSense’s Enterprise Intelligence. With AI agents, research and data analysis can be transformed even further, as agents autonomously retrieve, analyze, and synthesize vast amounts of information, enhancing productivity and decision-making.
Microsoft’s Magentic-One, an open-source “generalist” multi-agent framework designed to manage complex, multi-step tasks, features an ‘orchestrator’ agent that directs specialized agents — Websurfer, FileSurfer, Coder, and ComputerTerminal — to enhance productivity and efficiency in tasks such as data analysis and information retrieval.
In January, Cohere unveiled its new agentic AI offering, North, a low code platform for enterprises to build and deploy agents to help find information across global knowledge repositories, conduct research and analysis, and perform complex tasks spanning previously disconnected tools. That same month, Capital One launched an AI agent that helps customers buy a car, from researching and comparing vehicles to even scheduling test drives.
In a recent paper, scientists at AMD and Johns Hopkins University described how an AI agent acted as a research assistant, researching and designing an experiment, and then also conducting it and compiling the results. In early February, OpenAI unveiled deep research, an agent that uses reasoning to synthesize large amounts of online information and complete multi-step research tasks. Google Gemini released something similar (and with the same name) earlier in December 2024.
Reducing Technical Debt while Accelerating Software Development and Cybersecurity
GenAI is already enhancing software development and cybersecurity. When it comes to software development, AI agents will go beyond generating code — they will test, debug, and execute it, streamlining the development process and reducing errors.
More than 70% of the software used by Fortune 5000 companies was developed at least 20 years ago. AI agents will rewrite legacy code, helping reduce technical debt. A banking company looking to modernize 20,000 lines of code estimated it would need 700 to 800 hours to complete the migration; their genAI agent approach cut that estimate by 40%.
When it comes to cybersecurity, AI agents are set to revolutionize how threats are detected and mitigated. In December 2024, Fujitsu announced their multi-AI agent security technology, which coordinates multiple AI agents with different specialties to simulate cyberattacks, protection strategies, and business continuity measures.
Risks of Deploying AI Agents
While agents promise significant benefits, they also come with risks. Agentic AI requires high levels of trust from users, which, as one AI expert noted in an AlphaSense transcript, is a significant limitation in consumer applications. Although industry executives trust AI agents to an extent, 57% acknowledge the need for robust safeguards.
Financial Loss and Brand Damage
Poorly trained agents might make decisions that conflict with business objectives or ethics. In some cases, they could make a mistake that leads to financial losses or a damaged brand reputation. An engineering AI agent, if not properly managed, could lead to system outages or software bugs. In high-stakes environments such as healthcare or finance, even small errors can lead to significant consequences.
Security Vulnerabilities
Just like any other AI system, AI agents are vulnerable to those with malicious intent. Agents could be targeted by hackers, potentially exposing sensitive data or disrupting operations. Agentic AI systems can be hijacked to orchestrate harmful decision outcomes. This could lead to legal issues, customer distrust, and reputational damage. Coding errors within AI agents can also lead to unintended data breaches or security threats.
Overreliance on Agents
Heavy dependence on agents could also be a challenge, as it could result in a loss of human skills in certain areas. Also, as people become accustomed to AI agents making decisions, they may struggle to make choices without AI assistance. There is also a tendency to excessively trust AI systems — automation bias — which can result in acceptance of AI-generated output without the requisite verification.
From Experimentation to Mainstream Adoption
AI agents are shifting from experimental concepts to tools that drive real-world impact. From creative fields and regulatory compliance, to personalized healthcare and large-scale infrastructure management, AI agents will play a growing role in shaping how work gets done. As AI agent capabilities mature, businesses must evolve alongside them, which means investing in robust AI strategies while addressing the challenges that come with AI agents.
Discover how you can stay ahead of market-moving trends and transform your research process with AlphaSense’s industry-leading Generative Search.
Start your free trial of AlphaSense today.
ABOUT THE AUTHOR
Sarah Hoffman
Director of Research, AI
Sarah Hoffman is Director of Research, AI at AlphaSense, where she explores artificial intelligence trends that will matter most to AlphaSense’s customers. Previously, Sarah was Vice President of AI and ML Research for Fidelity Investments, led FactSet’s ML and Language Technology team and worked as an Information Technology Analyst at Lehman Brothers. With a career spanning two decades in AI, ML, natural language processing, and other technologies, Sarah’s expertise has been featured in The Wall Street Journal, CNBC, VentureBeat and on Bloomberg TV, and she was named a Rising Star for WatersTechnology’s Women in Technology and Data Awards in 2018. Sarah holds a master’s degree from Columbia University in computer science with a focus on natural language processing, and a B.B.A. from Baruch College in computer information systems. Sarah is based in New York.
Read all posts written by Sarah Hoffman
