Opinion: Why AI Agents Are The Future Of Everyday AI

2025 seems to be the year of Agentic AI. OpenAI, Microsoft, Anthropic, and other major tech companies have released their most advanced AI models yet, capable of performing complex tasks that weren’t possible just last year. But what exactly are AI agents, and how much of this technology is real versus hype?

Keeping up with AI trends in recent years can be exhausting. We’ve gone from basic AI to generative AI, then to reasoning AI, and now, AI agents are dominating the conversation. In the past few months, multiple tech companies have been introducing new AI agents and highlighting how workflows that connect them are becoming increasingly appealing.

In January, OpenAI launched its first AI agent, Operator, capable of performing tasks autonomously, such as browsing the web, booking a restaurant, or securing concert tickets for you, based on your preferences and criteria.

Two months later, the Chinese AI startup Zhipu released AutoGLM Rumination, a free AI agent that can operate independently and plan travel for users. And just a few days ago, Microsoft introduced “Computer Use” for Copilot Studio,  adding more autonomous capabilities to AI agents, including the ability to take control of the user’s desktop.

But this trend isn’t just about companies launching cool new products—it’s about a complex and evolving field that blends various forms of AI with human interaction. Tech enthusiasts and experts alike have shown strong optimism about Agentic AI.

“The one trend I am also excited about is AI agents workflow,” said AI expert and Coursera and DeepLearning cofounder Andrew Ng, at Snowflake’s BUILD 2024 conference in November. “If you were to ask, what’s the most important AI technology to pay attention to? I would say it is Agentic AI.”

So, what exactly are AI agents? Do they perform as well as the experts claim? Here’s a brief overview of the most essential aspects of this emerging technology and the latest developments.

What Are AI Agents?

Just as companies like OpenAI and Microsoft have shown, AI agents are technologies capable of performing complex tasks by autonomously combining different applications and software.

Essentially, they are “traditional” AI chatbots—but more advanced, capable of doing tasks beyond what we’ve been used to over the past two years. These include planning, browsing the web, remembering your interests, and even doing your grocery shopping without step-by-step guidance, as they figure things out on their own.

Google Cloud defines AI agents as “software systems that use AI to pursue goals and complete tasks on behalf of users.”

What makes these AI models truly special is their advanced reasoning capabilities, which allow them to process multiple types of information—audio, text, video, code, and images—simultaneously. They can understand complex situations the way humans would, take action, and interact with software and platforms.

Many tech companies are now pushing their AI models to this next level.

Just a few days ago, OpenAI announced that its AI models o3 and o4-mini can “think” with images and manipulate them—cropping or transforming them, for example. Meanwhile, Microsoft’s new “Computer Use” feature enables its AI agents to recognize graphical user interfaces.

“If a person can use the app, the agent can too,” said Charles Lamanna, Corporate Vice President, Business & Industry Copilot at Microsoft, in the announcement.

The Next Level: Agentic AI Technology

While many tech companies are developing ultra-smart AI chatbots, agentic AI goes a step further. This technology focuses on building systems in which multiple of these AI agents can interact and perform different tasks to deliver more accurate and complete results.

During her talk on “The Power of AI Agents” at a recent Tech Talk event I attended, organized by eDreams, AI expert Patricia Gutierrez Faxas explained that these systems can take on various structures. Multi-agent systems can break complex tasks into simpler ones, assign them to specialized agents, and form different types of networks.

An AI agent can take on a “Supervisor” role, delegating smaller tasks to other specialized agents, or multiple agents can work simultaneously on separate tasks and coordinate to deliver the final output. Some of the benefits of these systems include improved accuracy, greater privacy, and reduced errors.

Andrew Ng agrees with this approach. During his talk at BUILD 2024, he emphasized the importance of Agentic AI workflows for processing complex tasks with much better outcomes. The leap in performance from models like GPT-3.5 to GPT-4—and their stronger results on coding benchmarks—is largely due to agentic workflows and agentic reasoning.

How Will AI Agents Affect Our Lives?

Tech companies are taking different approaches to Agentic AI. OpenAI has focused on ideas for everyday users. When launching Operator, the team explained in a video that a person could simply think of a dish they’d like to prepare, ask ChatGPT to find the best recipe, and even have it purchase the necessary ingredients from an online supermarket.

The AI model could ask follow-up questions and pause before executing critical tasks, like completing a purchase using the user’s credit card. Users can set spending limits or require the assistant to ask for confirmation before buying anything.

The possibilities for how regular people can use AI agents seem endless, but the most popular use cases so far include acting as travel agents, personal shoppers, and administrative assistants—after all, who wouldn’t want a personalized technology that sends emails, remembers important birthdays, and takes care of taxes?

Microsoft has taken a more corporate approach. The company has demonstrated how marketing agencies can use AI agents to automate market data collection and generate insight-driven analyses, or how large enterprises can use them for invoice processing within their systems.

Still, while all of these use cases sound revolutionary, experts agree that there’s a long road ahead. Most people can’t yet use AI agents to improve their daily lives, as many of these features, like Operator, are only available to Pro users, a subscription that costs $200 per month.

So, Is Agentic AI Matching The Hype?

Many media headlines are promising a new AI revolution led by Agentic AI. There’s certainly a lot of hype in the industry—and the reality may not be far behind. A recent report shared by IBM revealed that many developers are actively exploring and developing this technology.

“IBM and Morning Consult did a survey of 1,000 developers who are building AI applications for enterprise, and 99% of them said they are exploring or developing AI agents,” explained PhD Maryam Ashoori, Director of Product Management at IBM watsonx.ai. “So yes, the answer is that 2025 is going to be the year of the agent.”

However, experts note that there are multiple interpretations of what AI agents truly are and how they should perform. As a result, the progress and performance of current AI agents often depend on the perspective and standards of the person evaluating the technology.

What is certain is that many experts and decision makers deeply believe in this technology and are investing heavily in its development. A recent study revealed that 84% of IT leaders trust AI agents as much as humans, and OpenAI expects around $125 billion in revenue by 2029.

Although Agentic AI technology is not yet fully developed and its adoption isn’t as widespread as that of chatbots—largely due to its high cost—it’s clear that the overwhelming optimism and enthusiasm around this trend are powerful enough to drive major changes in the months ahead.