How an IT Company Increased AI Agent Usage from 15% to 70%: Overcoming Team Resistance

In early 2026, despite AI agent technology being stable for six months, a survey at a large IT company revealed that only 3 out of 20 ML specialists, who develop medical image pathology detection systems, had tried using AI agents in their work. This meant only 15% of employees actively used the new technology. Within six months, the company managed to raise this percentage to 70% by adopting a systematic approach to overcoming resistance.

Implementing any new technology, even one promising to simplify life, invariably encounters human factors: resistance to change, inertia, and an unwillingness to step outside comfort zones. This is particularly evident in highly skilled teams, where each employee values their expertise and established processes. This slows down innovation, blocks productivity growth, and leads to a loss of competitive advantage. However, this is solvable, and there are proven methods to gently yet effectively guide a team through this transition.

Why Even AI Specialists Resisted AI Agents

The problem wasn't a lack of knowledge or qualifications. The team consisted of hardcore ML specialists who deeply understood models and their inner workings. They grasped the potential perfectly but were slow to apply it in practice. The core reason lay in human psychology: people are reluctant to disrupt their routines, even if those routines are inefficient.

Management faced a paradox: a team that built cutting-edge AI solutions was hesitant to adopt them in their own work. This led to missed opportunities, slowed internal processes, and underutilized valuable intellectual potential.

The Path to AI Agents: From Observation to Targeted Implementation

Initially, the company didn't impose top-down changes but opted for observation. Leaders monitored how the team adapted to new challenges and what tools emerged organically. This approach proved crucial: it helped identify real "pain points" and show where AI agents could deliver maximum benefit.

For example, when a manager went on vacation, the team faced an increased workload. During this period, behavioral changes became noticeable: some took on organizational roles, others tackled new tasks. It became clear that employees were actively seeking optimization methods. It was also observed that each team began developing its own internal data tools, a function previously handled by a separate department. This suggested that AI agents could significantly accelerate the development of such solutions.

Designing the Implementation: "The Elephant, the Rider, and the Path"

The implementation strategy was based on the "Elephant, Rider, and Path" metaphor from the book "Switch". The Elephant represents the emotional brain, the Rider the rational brain, and the Path the environment. Successful implementation requires addressing all three components. This meant not just providing a tool, but creating conditions where its use became natural and desired.

For the "Elephant" (emotions):

• Visibility of success. Early adopters were made prominent, and a dedicated space was created for sharing experiences, questions, and news about AI agents. This fostered a "everyone's using it, maybe I should too" effect.
• Inspiring examples. Particularly compelling was an AI-powered tool for blind code comparison. Its success impressed a team lead, who actively integrated agents into her team, making it a leader in AI usage within the company.

For the "Rider" (reason):

• Concrete, not abstract goals. An initial mistake was giving team leads abstract goals, leading to endless planning. Later, the approach shifted: they started with one standard task that everyone had to perform using AI agents, such as code review. This gave employees immediate practical experience.
• Training and workshops. Workshops were held on how to prompt AI agents, and documentation was updated so agents could access it.

For the "Path" (environment):

• Barrier removal. The company identified and eliminated real obstacles: VPN issues, insufficient disk space, complex instructions. Special sessions were organized for AI agent setup, where employees received technical support and resolved their problems on the spot.
• Incentives. A dedicated day for AI agent setup, ending with an early finish, provided an additional incentive.

Implementation and Adaptation: From Innovators to Skeptics

Implementation occurred in stages, considering different employee groups based on Rogers' Diffusion of Innovations model (innovators, early adopters, early majority, late majority, laggards). Each group required specific arguments and approaches.

• Early Adopters (13.5%). These employees actively engaged, wrote use cases, and voluntarily participated in pilot groups. Demonstrating the innovation's value for practical problem-solving was sufficient for them.
• Early Majority (34%). This pragmatic group required proof of success. They were shown cases from early adopters and testimonials from trusted colleagues. The example of the team lead, inspired by the tool's demonstration, illustrated this approach's effectiveness.
• Late Majority (34%). Skeptics who were risk-averse. For them, "out-of-the-box" solutions were created (e.g., a proxy server for VPN-free access), and training was provided before use.
• Laggards (16%). This group is the least willing to adopt change. The decision was made not to force adoption immediately but to wait until AI agents became mandatory in processes where they delivered significant impact (e.g., incident analysis).

One of the most striking examples of successful adaptation was the creation of a tool for 3D image annotation. Previously, this process took up to 10 hours per image, deterring medical professionals. With AI agents, the development of such a tool took only two weeks, and the annotation process resumed.

Results

Within six months, the company increased the percentage of active AI agent users from 15% to 70%. This led to a significant acceleration in the development of internal tools and an overall increase in efficiency. For example, the time to create a 3D image annotation tool using AI agents was reduced by more than 6 times, allowing a crucial process, previously abandoned due to its labor-intensive nature, to be reinstated.

How to Implement This in Your Company

Overcoming resistance to change is not a one-time event but a systematic process. If you are facing a similar situation, here's where to start:

• Observe and identify real needs. Don't impose the tool; rather, see where people are already looking for solutions. This will highlight the most promising areas for AI agent application.
• Create inspiring examples. Make early adopters visible, share their successes, and demonstrate how AI agents solve specific problems. Emotions are a powerful driver.
• Set concrete tasks. Avoid abstract goals. Start with one standard task that everyone must perform using AI agents. This provides immediate practical experience.
• Remove barriers. Conduct surveys or one-on-one conversations to understand what prevents employees from using the tool. Address real problems, not just those articulated on the surface.
• Adapt your approach to different groups. Remember that innovators, early adopters, early and late majority, and laggards require different arguments and implementation strategies.

If this case sounds like what's happening in your company, our manager can help: he'll analyze your business and niche for free and point out where an AI agent would bring a real result in your case. Message the manager