What opportunities does Agentic AI unlock for CVM?

Whether generative, predictive, or agentic, AI doesn't change the fundamental challenges of CVM in telecom: such as maintaining ARPU, reducing churn, personalizing offers, optimizing marketing efforts, and maximising customer value.

However, in markets characterized by customer volatility, multi-SIM usage, price pressure, and the complexity of channel orchestration, it transforms the speed, precision, and scale at which teams can operate.

Applied to CVM, AI can become an engine of value creation across the entire customer base, even more so when combined with an AI agent able to analyze, recommend, arbitrate and orchestrate workflows simultaneously to bring the best opportunities to light.

1. As CMO, what is your top priority today in customer marketing and CVM?

My absolute priority is to close the "speed gap" between generating insights and our operational Go-To-Market (GTM) capability.

Historically, CVM relied on retrospective analyses and aggregated segmentation. Today, our priority is to shift to real-time hyper-personalization, driven by high-frequency data. It's no longer just about reacting to customer behavior, but about getting ahead of their needs by identifying micro-niches of value that traditional tools can't see.

At the same time, we put inclusion and empathy at the heart of our strategy (as with our #MyCustomerMyBoss initiative). My role as CMO is to make sure algorithmic acceleration serves a universal customer experience, without ever losing touch with the reality on the ground.

2. Telecom markets in MEA are marked by fierce competition, multi-SIM usage and price sensitivity. Why is AI particularly relevant here?

These specificities make the MEA market a highly volatile environment where the customer's decision is made in a split second, with multi-SIM usage, constant price arbitrage and an unpredictable regulator on top.

• Detecting "value gaps": just as cybersecurity agents hunt for system vulnerabilities, AI can continuously scan our usage data to spot attractiveness gaps and micro-signals of churn well before a customer switches SIM cards.
• Dynamic price elasticity: price sensitivity doesn't mean "always cheaper", but "the right value at the right time". AI lets us simulate and dynamically test price sensitivity so we can offer top-ups, bundles or Mobile Money services perfectly matched to the user's immediate purchasing power. A/B tests can come to life and steer commercial activity.
• Excellence in action at scale: individually managing millions of prepaid customers with highly erratic usage behavior is humanly impossible. AI brings the scale that marketers have long wished for.

3. What does agentic AI bring to CVM in terms of personalization compared to traditional NBO or churn prediction?

Traditional CVM (NBO, churn scoring) relies on predefined business rules or passive statistical models: if the customer does X, we trigger Y.

Agentic AI changes the game in three major ways:

1. Autonomy and proactivity: the AI agent doesn't just execute a score; it assesses its environment, tests hypotheses, orchestrates scenarios and takes initiatives adjusted in real time (or near real time, depending on data availability). Of course, this has to happen within a controlled framework, otherwise its own logic could lead it to cut prices by 99% to maximize customer satisfaction.
2. From segmentation to nano-segmentation: we go beyond targeting by "cohorts". The AI agent interacts directly with the data to isolate unique usage behaviors at the individual level.
3. Continuous experimentation (dynamic A/B testing): instead of launching a campaign and measuring its impact a month later, the AI agent continuously adjusts the offer, channel and timing to maximize customer value and retention rate.

4. What kind of copilot model should we imagine between CVM teams and AI agents?

Two pitfalls must be avoided: "blind autopilot" and using AI as nothing more than a sophisticated search engine.

The right copilot model is an augmented collaboration where the marketing manager and CVM teams become the architects of the algorithms:

• The AI agent handles the analytical complexity, which includes data exploration, detection of hidden patterns, scenario simulation and execution of micro-campaigns.
• Humans bring strategic vision, empathy and governance: our CVM teams define the ethical guardrails, validate value alignment and inject the commercial intuition that data alone can't guess.
• The marketing manager factors in constraints such as regulatory compliance (price floors or ceilings), budget limits and ethical considerations.

The AI agent proposes and simulates; the marketer validates, refines and sets the strategic direction. This requires updating operational processes so that the time saved on the analytical side isn't lost.

5. You've just gained Early Access to our customer intelligence agent. What are the first benefits you're seeing?

The first benefits are striking and directly affect our operations.

• First, I see a spectacular gain in speed of access to information. Usage or cohort analyses that used to take several days of SQL extraction and Data Lab querying are now available in seconds through natural language.
• Second, it frees up creative time for our Data Scientists and Marketers. Less time spent preparing data, more time devoted to designing offers and GTM strategy.
• Finally, a new view of micro-niches. The agent brings to light ARPU pockets and usage segments overlooked by our traditional dashboards. It's worth noting that our current analyses, shaped by our own preconceptions, didn't allow us to discover all the viable and profitable niches.

6. Beyond technology, what is the most difficult to change: tools, data, processes, or culture?

Without hesitation: culture and processes.

Technology progresses exponentially, but organizations evolve linearly. Increasing the speed of access to information through agentic AI is pointless if the decision chain to validate a product or launch a campaign takes three weeks.

• Data quality is a major technical prerequisite that requires us to rethink our pipelines and enrich our databases. The useful knowledge generated by the agent must also enrich our customer databases to serve the front office and back office during customer interactions.
• But the culture of change is the real bottleneck. We need to move from a "monthly report" culture to a culture of permanent experimentation, reassure teams about the role of AI (which augments the marketer without replacing them) and adapt our operating model to the pace of the algorithms.

7. How do you see the agentic ecosystem evolving in telcos, with several business agents able to collaborate?

We are heading toward automated multi-agent orchestration that will redefine the very structure of our companies.

• Tomorrow, a marketing agent will detect a niche of value and design an offer.
• It will collaborate in real time with a customer service agent to adapt response scripts and anticipate incoming requests.
• It will consult a business & financial risk agent to validate margin and compliance, while relying on a data & network agent to check network capacity.

In this near future, the CMO's role is evolving into that of a CTO of customer experience and value architecture. The marketing leader will no longer be just a communicator or a creative, but the conductor of an ecosystem of autonomous agents serving inclusive growth and customer satisfaction.