From Co-Pilot to Co-Worker: Where the AI Assistant Journey is Headed to Next A new breed of autonomous AI is set to take 2025 by storm.

Given the current prevalence of artificial intelligence (AI), it's easy to forget that the technology has only recently moved from the research lab to real-world business applications. Despite the decades of development that have laid the groundwork for current AI solutions, we're still just scratching the surface.

Of the many AI applications, the rise of AI assistants has been on a particularly remarkable journey. We've seen the technology move from chatbots that were limited to simple decision trees, to sophisticated customer support tools that can handle all manner of queries. Almost 1.5 million people had at least one conversation with a chatbot within the past year.

More recent years have seen the growth of AI co-pilots, ready to assist on anything from producing code and managing calendars to searching for online materials. Yet even the capabilities of co-pilots, which are a phenomenon of this year, are already being superseded by AI agents.

These autonomous AI assistants go far beyond the traditional query-response capabilities of chatbots and co-pilots to deliver enterprise-grade outputs without guidance. In fact, the excitement around AI agents saw them top Gartner's list of 2025 tech trends.

Unsurprisingly, OpenAI is keen to make a play for its share of the AI agent market. However, its recently announced offering, called Operator, won't be coming to Europe anytime soon, according to CEO Sam Altman.

This isn't the first time that Big Tech has struggled to scale AI products in Europe. In 2024 Google, Meta, X and LinkedIn paused or delayed AI projects in the European Union, placing the blame on excessive privacy regulations and red tape.

Yet this doesn't mean AI agents are off the table here. Instead, we're seeing an uptick in innovation from European companies that also benefit from a much closer understanding of specific demands and opportunities that are much broader than data privacy regulations alone.

For example, Deutsche Telekom has been working to solve the challenge of scaling an intelligent operator that can work across borders and languages. Here, a multi-agent architecture and systems design to prepare for a rapid roll-out of AI agents across the 10 countries that make up its European footprint.

As the technology surrounding AI agents continues to mature, it's time to take a closer look at the technology, how best to approach early-stage adoption and get in front of data governance best practices.

AI agents as the new frontier

As the name suggests, AI co-pilots exist to support the goals and activities of the actual user. In contrast, AI agents represent a sizable leap forward as they have capabilities closer to a co-worker, with the ability to work intelligently, independently, and autonomously. These AI agents are autonomous intelligent systems performing specific tasks without human intervention.

The promise of AI agents sits with their ability to interact with the environment, collect data and use these findings to perform tasks that meet predetermined goals. For example, an AI agent may be leveraged to resolve customer queries in a contact center. Here, the tool would automatically ask the customer different questions, look up information in internal documents, and respond with a solution. Based on the customer responses, it determines if it can resolve the query itself or pass it on to a human. Yet AI agents can go far beyond customer service, which has been a typically popular use case for preexisting chatbot technology.

The new breed of autonomous AI is set to take 2025 by storm thanks to its impressive ability to deliver context-aware decision-making. With defined workflows and autonomous systems, AI agents can handle everything from environmental assessments to C-suite tasks, problem-solving at scale to providing creative solutions. While the user will set the goals, the new breed of AI agents are impressively independent, self-determining which tasks will help them reach said goals most effectively and efficiently.

For example in customer service, AI agents offer startups and enterprises new pricing models, such as conversation-based pricing, which align costs with work output and provide simplicity. This flexibility allows companies to iterate faster, giving them an edge over incumbents.

Similarly for B2B-focused companies in sales and marketing, marketers are transitioning to strategic roles with AI-driven automation, enabling hyper-personalization and seamless sales-marketing integration. However, trust, content quality, and brand consistency remain challenges, while AI offers opportunities for SMBs and new efficiencies.

Why autonomy matters: Driving competitive advantages in the enterprise

While an estimated 25% of enterprises are planning to deploy AI agent technology in 2025, this figure means that the vast majority of large organizations are yet to recognize the potential of this tool.

For decision-makers still on the fence, it's important to temper the benefits of AI agents such as reducing unnecessary costs arising from process inefficiencies, human errors, and manual processes against potential risk factors to move forward with the decision. For large enterprises, concerns typically relate to integration complexities and legacy dependencies. Internal data silos, ethical considerations and governance challenges.

These are all valid challenges that need a clear plan of action. For this reason, executive teams are the ideal candidates to spearhead the adoption of AI agents in 2025. This is because these individuals have the clearest insights into where cross-functional handoffs might create friction, the departments with the most pressing need, and the occasions where high-value activities may stall due to fragmentation.

A sensible starting point lies with high-performing employees with a strong mastery of cross-functional processes. These individuals can use these to create pilot projects that allow AI agents to learn how to handle complex organizational workflows and the tasks within that deliver toward the set goals.

Another sensible strategy is to deploy AI agents in clusters so that they can learn from each and make decisions that drive efficiency and productivity. It's also important to remember that AI agents don't exist in a vacuum. While they represent a leap forward from existing co-pilot tools, even the actions of a human co-worker need to consider colleagues and other contributions.

This means AI agent adoption needs to involve the employees who will work alongside these intelligent digital assistants to monitor progress, create performance feedback loops, reward good performance, and tackle friction points head-on.

With these pilot projects underway, enterprise organizations can move off the starting block in 2025 without delay and prove the business value of AI agents needed to justify further usage.

Improving AI agents with data and governance, and how leaders can enable autonomy

The demand for AI agents will rise further as the competitive advantage of their autonomous power becomes clear. As with all forms of AI, the capabilities of AI agents directly correlate to the data at its disposal. Although the excitement around AI means that two-thirds of the respondents to Forrester Research's 2024 State of AI Survey believe their organizations would require less than 50% return on investments to consider their AI initiatives successful, this percentage doesn't need to be the standard. While some degree of error has been accepted as standard within AI, there are strategies available to reduce the associated risks that pose a concern for business leaders.

Leaders need to agree in advance on quantifying what leaders' class as "good enough" in terms of results and agreeing on acceptable error rates. Research from Ness Digital Engineering found that 71% of high-performing systems for industry can achieve autonomy without compromising reliability. These results were thanks to clear requirements for the AI agents and a modular design system. [Editor's note: The author is the Chief Growth Officer at Ness Digital Engineering.]

This shows that enterprises don't need to strive for immediate perfection here. Moving beyond pilot projects toward organization-wide implementation can take a measured approach with both data and strategy.

Here, leaders can begin to implement AI agents in the places where data is the strongest. This is also true when it comes to deciding how to roll out the technology. These don't need to stretch across every system and process, especially not in the earliest stages. As with human co-workers, AI agents can be applied to the pressure points most in need of support or which deliver the highest-impact activities.

From here, the technology can be trained and optimized to perform autonomously within this area, rather than striving for enterprise-wide performance and excellence, which will require significant time and resource investment. With these hotspots clearly identified, guardrails and governance systems can then be applied to allow these agents to work autonomously with appropriate management as with any organizational employee.

A vision for 2025 and beyond

AI agents are not yet widespread, but 2025 is set to be the year they make a splash within specific verticals. Still, given the low penetration rates currently seen at complex enterprises, the decision to move forward with AI agents means a sizable competitive advantage is up for grabs. For European companies, innovative homegrown startups that address niche use cases provide a sensible entry point for the technology.

Successful integration will need a close understanding of operational systems and understanding of the high-impact data sources that will set AI agents up to deliver value in rapid time.