When AI starts making decisions: Rethinking the customer journey

For a long time, AI in marketing has been framed as a tool that helps humans make better decisions. But that framing is starting to break down. Because the real shift isn’t about assistance anymore—it’s about delegation.

When AI systems can make decisions in real time, at scale, across millions of interactions, the question is no longer how they support humans. It’s what happens when they start acting on their behalf.

For users, this changes the experience immediately. The journey stops feeling like a sequence of steps designed by a company and starts behaving like a conversation that adapts to them. Instead of navigating content, waiting for responses, or being routed between touchpoints, they interact with a system that understands intent, asks questions, and moves them forward instantly. The experience becomes fluid, continuous, and responsive—closer to speaking with a knowledgeable person than interacting with a digital interface.

But for organizations, the impact is deeper.

Because once AI takes over decision-making in high-volume, low-risk moments- answering questions, qualifying intent, recommending next actions, the role of humans has to change. These are not decisions that benefit from human involvement anymore. They happen too often, too quickly, and at an overwhelming scale. Keeping humans in that loop doesn’t add judgment—it adds latency.

At the same time, some decisions cannot be delegated.

In regulated or sensitive environments, the stakes are fundamentally different. You don’t want to be the CMO explaining to a regulator why symptoms weren’t described accurately in a healthcare interaction, or why financial advice generated by a system wasn’t compliant. In those moments, human judgment is not optional; it’s accountability. AI can assist, but it shouldn’t decide.

This creates a clear divide. AI owns the high-frequency, low-risk decisions that define most of the customer journey. Humans remain responsible for the high-stakes, high-risk ones where precision, compliance, and accountability matter most.

But even that framing is incomplete.

Because the most important shift is not just which decisions humans make, it’s where they operate in the system.

Humans are moving out of the loop of execution and into the layer of design. Instead of reviewing outputs, they define the system itself: what knowledge it has access to, how it interprets intent, what guardrails it follows, and what it is allowed to say or do. The system executes continuously, but humans shape its behavior.

This is where the idea of “human-in-the-loop” starts to evolve. In many cases, it no longer means approving decisions in real time. It means designing the conditions under which decisions are made. And that shift becomes even more visible when we look at how customer experience itself is changing.

We are moving away from a model of pushing content to segments, toward interacting with individuals in real time. In a conversational system, every interaction is different. The system builds context as it goes, asks questions, adapts its responses, and guides the user based on what they need.

This is the foundation of what we’re building at Kaltura.

By combining real-time conversational avatars with personalized content intelligence, the experience becomes inherently adaptive. The system doesn’t rely on predefined journeys or static assets. It pulls from across the organization’s knowledge, videos, demos, webinars, sales conversations, and uses that intelligence to respond, qualify, and guide in the moment.

In that sense, content, especially video, stops being a static output and becomes an active part of the decision-making system. It’s no longer something you distribute. It’s something the system can understand, break down, and use to move users forward.

But as AI takes on more of this role, another challenge emerges.

The system starts to feel like a black box.

You define an outcome, and it delivers results, but you don’t always understand how those results were achieved. Decisions happen in real time, across thousands of micro-interactions, making it difficult to trace cause and effect. Performance improves, but transparency decreases.

This is where many organizations risk losing control, not because the system fails, but because they can’t explain it. Overcoming this doesn’t mean limiting AI. It means changing how we measure and observe what it’s doing.

Traditional measurement was built around predefined journeys and clear attribution. But in a conversational, AI-driven environment, those models break down. Journeys are dynamic. Decisions are continuous. Outcomes are the result of accumulated interactions, not isolated events.

So, measurement must shift.

Instead of focusing only on outcomes, we need to understand decision signals. What did the user ask? Where did they hesitate? What objections surfaced? What moved them forward?

This is where Kaltura’s approach to content intelligence plays a critical role. By structuring and analyzing video-based interactions, sales calls, demos, and webinars, the system doesn’t just deliver answers; it captures and exposes the signals behind them. You can see patterns of intent, friction points, and moments of progress.

The black box doesn’t disappear, but it becomes observable.

And that fundamentally changes how organizations operate. Measurement becomes less about reporting performance after the fact, and more about understanding decisions as they happen.

In that world, the customer journey is no longer something you fully design in advance. It’s something that emerges in real time, shaped by the interaction between the user and the system.

Which leads to the final shift.

The companies that win won’t be the ones that simply add AI components into existing workflows. They’ll be the ones that redesign those workflows from the ground up—AI-first, not AI-assisted.

Because once AI takes over high-volume decisions, and humans focus on the high-stakes ones, the advantage shifts to those who know how to design for that reality from the start.