Episode #1 - TALK TO MY AGENTS !

The expansion of Agentic AI

By 2030, intelligent agents are projected to manage 30 to 40% of all digital interactions, fundamentally reshaping how businesses engage, support, and serve their customers. This shift, backed by multiple industry studies, marks a pivotal transformation in the digital landscape. This percentage will likely be higher for certain interactions such as e-commerce or customer service and support.  A recent report[i] shows that, in a B2B context, 68% of companies surveyed expect their interactions to be handled by agent AI in the next three years. 

So, what is agentic AI? These terms refer to autonomous artificial intelligence systems capable of setting goals, planning, and executing multi-step tasks with minimal human intervention, making independent decisions and adapting to dynamic environments. As we previously explained in a Talan blog article, it is not limited to simple AI agents[ii].

We are therefore on the cusp of a radical change in how individuals or citizens will establish their omnichannel relationships with brands, in B2C/B2B, and also with public institutions.

Inventing new experiences/interfaces

In practical terms, this means we’ll engage with entirely new interfaces that are intuitive, intelligent, and capable of handling tasks ranging from the mundane to the highly complex. These interactions will be multimodal, seamlessly blending vision, voice, and text to create fluid, human-like exchanges. Autonomous agents won’t just respond, they’ll collaborate, self-organise, and drive toward the goals we define, transforming how we work, communicate, and live.

This is what Demis Hassabis, co-founder & CEO of Google DeepMind and Nobel Prize winner[iii], describes: we will increasingly interact with less primitive forms of interfaces than those we know today, in the form of prompts. They will be much more multimodal, combining multiple forms of communication. In this regard, he readily cites Steve Jobs and the three key dimensions of a successful experience: simplicity, beauty, and elegance. The complexity of cutting-edge AI must be masked to unleash its full potential – the future of AI will be defined not just by intelligence, but how gracefully it integrates into our lives. 

Therefore, modules of these agents will inherit specialised tasks. They are currently called tools (libraries of tools available to an AI agent). They will interact with their traditional or agentic ecosystem to achieve their objective. Of course, as time goes on, they will prefer to communicate with other agents or tools that will allow them to fulfill their mission more effectively. Interoperability protocols are emerging, such as the agent-to-agent protocol[iv] or the Model Context Protocol. The issue of trust and security will be decisive. The agentic world will have to be built on these foundations. Otherwise, it will not come into being.

The battle has already begun to control the future "gateways" of agentic interactions. For illustration, this year, Perplexity launched the Comet browser, Google announced the disappearance of its famous blue links in favor of a more agentic experience, Microsoft, Salesforce, and ServiceNow are integrating AI agents into their services... and the list continues to grow, for humans and for machines.

What's at stake for brands

This is the tipping point for brands. Those that fail to anticipate and adapt to the emerging landscape of multimodal, agent-driven interactions risk losing their grip on customer relationships entirely. The ability to operationalise this new paradigm isn’t optional; it’s a strategic imperative. Brands that move decisively will redefine customer experience and extend their influence. Those that hesitate may find themselves sidelined, watching loyalty and engagement slip away.

For example, I could delegate to my agent the organisation of my weekend in a modern-style hotel, near the train station because I will be traveling by train, within my price range, and with certain types of amenities... this simple sentence implies a lot: what is my definition of "modern style"? of "proximity"? etc. Over time, my agents will learn what this means to me and remember it. It will no longer be me but them who will make the selection or pre-selection. This is what underlies the promise of hyper-personalisation.

For public institutions, less subject to a competitive environment, the challenge will be to maintain or develop their relationship with citizens through a new kind of public agent.

From the brand's perspective, this means rethinking its traditional operations to become visible: paid inserts, attractiveness to search engines (SEO/SEM), websites, applications, and induced visibility on social networks like TikTok or Instagram. In itself, an agent will not "see," in the literal sense of the term, these devices. They will therefore be useless because they are designed for human eyes. We have seen the emergence of GEO (Generative Engine Optimization), but this approach remains limited given the global challenges.

How should brands organise themselves? 

Brands must develop and implement a truly visionary strategy because the speed of AI progress will make possible in six months what is not possible today. At the same time, they must also remain very pragmatic and show concrete short-term results. This is all the more important as we are in an unstable state and many parameters will still change. Depending on its tactic, a brand can be either a pioneer or a follower on certain topics, by choice.

For the brand, a first phase will consist of concretely projecting itself into a mixed digital world, both traditional and agentic. It must review all of its omnichannel customer/prospect relationships and consider the practical implications of this dual mode. This reasoned mapping will show the areas of the value chain that present the best potential: impact – risk – adoption – feasibility. For example, an AI agent could be responsible for streamlining traditionally disjointed and frustrating communications between the company, the carrier, and the end customer, particularly to proactively manage delays or changes in delivery locations.

In a second phase, it can define how this new approach influences its operational model and what needs to be changed.

It can then take action according to a plan built around its specific challenges. It will progressively integrate into the agentic ecosystem and structure its pilot projects. It can deploy its first externally visible agents according to the main protocols and iterate, while verifying its ability to technically demonstrate that it is trustworthy and secure. Finally, it must be able to quickly scale up if it observes strong customer adoption of one of its agentic services. This requires real-time collection and analysis of data that reflects the evolving agentic experience.

None of our digital interactions will escape this paradigm shift. And if we also imagine that AGI (Artificial General Intelligence) has a 50% chance of emerging by 2030 and that "the impact will be 10 times greater than the Industrial Revolution and will arrive 10 times faster," again according to Demis Hassabis, it is important to actively embrace this dynamic to avoid being left behind. 

Brands must therefore approach this new era with mastery, speed, and responsibility. The role of humans will remain decisive because it is they who will shape a new customer experience, making it unique. This fantastic challenge involves developing a very precise understanding of agentic AI, its potential, its technologies, and its risks, to imagine new universes that will resonate with other people while building strong links with AI agents in the brand's ecosystem.