By Yaëlle Philippe-Auguste, Director, Marketing, eStruxture Data Centers

I recently attended a conference focused on agentic AI. It was the kind of experience that reminds me why I am so passionate about this field. I get to learn, explore how AI is evolving, and connect new ideas to both my professional work and my everyday life.

Reflections on AI Momentum in Canada

The event opened with comments from a former federal minister who shared her impressions of the November 4 budget announcement. She highlighted a growing intention, from her perspective, to collaborate more actively with Canadian companies, accelerate AI development, and retain both intellectual property and AI talent within the country.

She was clear that these were her interpretations of the direction the government may take. Still, hearing this perspective felt both relevant and encouraging. At eStruxture, we are fully Canadian and building the largest data center platform in the country. Our recent partnership with Hypertec, ThinkOn and Aptum aims to deliver an end to end sovereign government solution, and it aligns naturally with the broader conversation happening at the national level.

Designing Agentic Applications in a Practical Way

One of the presentations focused on how to design effective agentic applications. The speaker broke down the reality behind the technology and shared several practical observations.

Here are the ideas that stayed with me:

• Generative AI is currently the easiest technology for organizations to adopt, and return on investment often comes quickly.
• Building an agent is challenging, although the biggest failure point usually comes from poor task definition, not the technology itself.
• An agent does not always benefit from a mission that is too narrow. If the objective is overly restrictive, the agent loses the ability to plan, reason and adapt. A clearly defined goal with room for decision making tends to perform better.
• The concept of deep agents was highlighted. These systems can break work into sub tasks, delegate to other agents, keep notes, and reference the files they create. This reflects current research and the evolution of multi agent systems.
• The speaker suggested that by 2028, agents may be able to autonomously execute tasks that span several weeks. This was presented as a forward looking projection, not as a confirmed fact.

This session helped translate the complexity of agentic AI into a set of practical design principles.

Understanding Agentic AI Compared to Traditional ML

Another presentation focused on the foundational differences between traditional machine learning and agentic AI. The distinction was simple but important.

Traditional ML predicts outcomes.

Agentic AI senses, reasons, acts, learns and adapts.

This shift changes how we think about intelligent systems. Instead of generating a prediction, an agent can plan a sequence of actions, use tools, interpret feedback and adjust its behavior. It can operate within an environment rather than simply output a probability score.

A Look at Agents in Action

The event concluded with a live demonstration of an agent interacting with unstructured data. Multiple PDFs were processed through a retrieval pipeline, and the agent responded to a simple query by synthesizing information across the documents.

The interaction felt natural and intuitive. Behind the scenes, the technology is complex, but the user experience is becoming more straightforward. This gap between complexity and usability is one of the reasons agentic AI feels so transformative.

How I Integrate AI Into My Work

What I learned today aligns with how I bring AI into my organization. I prefer to introduce AI progressively and with intention. I start with real business needs, and I aim for solutions that create measurable impact.

Here is what this looks like in practice:

• Improving processes that slow teams down
• Reducing friction and addressing operational bottlenecks
• Increasing rigor in execution
• Creating a lift in performance that is visible and sustainable

I have applied this approach in marketing, in sales, and across other teams. It helps us avoid pilot projects that never scale and instead build AI solutions that deliver value.

My Takeaway

Agentic AI is not about replacing people. It is about expanding what teams can accomplish. It combines human judgment with intelligent systems that adapt and take action. When applied with clarity and purpose, it becomes a multiplier of performance and potential.

This event made that future feel closer and much more tangible.