Agentic AI in Telecom: Field Notes From MWC Barcelona
- Maria H. Blake
- 6 hours ago
- 4 min read
XME.digital just finished our fourth MWC Barcelona, and our second year as exhibitors. The trip was intense and full of industry insights.
It’s impossible to ignore that Agentic AI took up half of the core agenda. We wanted to share the context gathered there and explain how the technology hits telecom and related markets. Read our notes and insights straight from the floor.
The Next Operating System of Telecom Is AI
AI moves into the core of telecom infrastructure to function as a network operating system.
Generative AI for basic support has become a commodity. Dr. Esteve Almirall notes that simple applications now create competitive parity rather than differentiation. Operators move beyond basic use cases and accept governance risks or "shadow AI" to solve a more urgent problem: managing complex infrastructure without inflating operations teams.
GSMA confirms this shift in priorities. Internal operational efficiency now drives up to 91% of AI initiatives. Service providers ditch experimental projects to build a base for future autonomous systems. Current deployments target three technical pillars:
Automated fault resolution (54%)
Customer experience prediction (52%)
Predictive network health scoring (48%)
Instead of investing in experimental projects, service providers lay the groundwork for future autonomous systems.
By 2025, over two-thirds of operators deployed Level 2 (L2) Autonomous Networks. These systems analyze events and suggest fixes, but humans still hit the "confirm" button. While the industry aims for L4 autonomy by 2030, messy legacy data prevent rapid progress. To fix this bottleneck, TM Forum, Huawei, Telefónica, and Orange launched the "L4 is ON" initiative. The group focuses on standardizing AI-ready data layers to move beyond human-in-the-loop decision-making.
"AI-driven infrastructure is the next logical step in telecom evolution. Modern networks have reached a level of complexity where real-time human management is no longer physically possible. Operators must restructure their processes to delegate operational decision-making to the system. This transition is far more challenging for the industry. We move toward a hybrid model where AI manages the network in real-time, while humans manage the AI systems themselves."
Victor Kmita, CEO, XME.digital
What Agentic AI Will Change
BI tools are the first in line
Agentic AI can radically change the analytics software market, just as SaaS once changed enterprise systems.
Providers move beyond experiments and integrate agents into closed-loop traffic management. The algorithm identifies an anomaly or segment overload and reconfigures the network in milliseconds. Users never feel a drop in speed, while engineers receive a report on a resolved issue.
The industry is ditching the old manual flow. Previously, systems collected raw data, tools generated insights, and a human studied a dashboard before making a call. Agents eliminate the "thinking delay." The chain shrinks to a direct result where data triggers immediate execution.
Agentic autonomy disrupts the established BI market. Dashboards lose their status as decision-making hubs. Interfaces now function as control panels or audit logs. An operator checks the screen not to find a solution, but to verify how the AI redistributed the load or optimized power consumption across the infrastructure.
New Operational Models Required
Agentic AI adoption fails when treated as a plugin for legacy environments. Rigid business processes and outdated infrastructure block the algorithm's ability to act. Service providers now rebuild organizational structures to eliminate bottlenecks. The transformation moves beyond headcount and hits the core IT infrastructure layer.
Example
Klarna deployed a customer service agent built together with OpenAI that performs the work of 700 human agents. Customer requests are resolved 81.8% faster, with higher accuracy and improved customer satisfaction. The automation adds around $40 million in annual profit.
Klarna has also announced that it’s moving away from Salesforce and Workday, and building its own agent-based operating system. They want to avoid one-size-fits-all SaaS workflows and operate with a custom operational model designed around AI agents. Klarna replaces legacy systems and rebuilds operations around the capabilities of AI agents.
Intent-Driven Interface
Legacy interface models rely on navigation. Users click through menus and forms to trigger functions. Intent-driven execution replaces the manual flow. A user expresses a goal, and the system orchestrates the necessary services to achieve the result.
Modern products no longer target human users exclusively. Front-end architecture must now support agentic interaction, ensuring that AI can interpret elements and execute tasks without human intervention.
This year, our delegation showcased a new vision on how a customer portal interface should be built to satisfy demand for an intuitive and fast customer cabinet and protect telecom service providers from AI ‘black box’ risks. On top of the XME Digital Service Platform we designed Stella#5, a generative customer portal that creates intent-driven interfaces using pre-configured widgets.
Book a demo to see Stella#5 generate customer interfaces in real time |
AI Returns Telecom to the Center of the Tech Ecosystem
The direction of venture capital suggests that the next wave of innovation will come from the intersection of telco and AI.
GSMA and Wing VC launched the Nova Future Summit at MWC Barcelona. The summit provides a closed platform where AI and fintech startups access operator infrastructure and established customer bases. Investors return to the telecom sector to find growth in new markets. Capital flows toward AI infrastructure, quantum networks, telco security, and autonomous agents.
Operators build internal investment mechanisms to maintain control over the ecosystem. Newly formed corporate venture units fund startups capable of expanding the core business. These investments target mobile financial services, cloud gaming, and AI-driven products. Telecom assets serve as a launchpad for technology markets. The industry moves from providing connectivity to hosting entire digital economies.
Comments