Agentic AI starter: the revolution in IoT fleet supervision

With the rise of the Internet of Things (IoT), fleet managers now have to oversee tens, or even hundreds of thousands, of devices. Detecting incidents, prioritizing interventions, and avoiding costly visits pose significant challenges. Additionally, the vast amount of data generated often requires advanced technical skills to cross-reference and analyze effectively.

The promise of AI agents: automating fleet management with artificial intelligence

Orange offers an innovative solution: IoT Agentic AI Starter. By combining Large Language Models (LLMs) with Orange’s Network APIs, this platform enables fleet managers to query their connected device fleet using natural language. No more complex interfaces or technical queries — simply ask a straightforward question to receive precise diagnostics and automatable actions.

How does it work ?

• Natural language querying: The manager asks a question or requests information, such as “Which objects are near my location?” or “What is the battery level of this device?”.
• Use of MCP tools: MCP provides tools that encapsulate Orange’s IoT platform functionalities, allowing secure access to data without technical complexity.
• Automation and proactive supervision: In the future, the agent will not only diagnose but also, based on predefined rules, propose or execute corrective actions. All automated actions are tracked and supervised by a human operator to ensure relevance and traceability of decisions.

A concrete example

During Orange OpenTech 2025, a demonstration using anonymized data from a real client of the platform illustrated this approach: a fleet manager of approximately 100,000 devices was able, in just seconds, to locate objects near Châtillon, check their battery levels, or diagnose those that no longer communicate. These actions, performed via simple natural language queries, showcase the power of this solution to optimize IoT fleet management.

Benefits for fleet managers

• Faster decisions: No need to manually cross-reference complex data.
• Reduced physical interventions: Automated diagnostics and corrective actions.
• Improved service quality: Increased responsiveness and proactive supervision.

A scalable and secure platform

Orange goes beyond mere connectivity. Customers will be able to interact with IoT services via natural language and automate fleet management through intelligent agents, while benefiting from guarantees on data security and confidentiality (encryption, access control, audit logs, compliance, and data localization). Orange will provide MCP servers and leverage its expertise to develop, deploy, and operate these agents.

Future perspectives

The architecture of Agentic AI Starter allows for expanding its applications to industrial automation, video surveillance, streaming, and automotive sectors. Orange aims to offer a flexible, scalable platform focused on simplification and security, to accelerate the adoption of new use cases and enable rapid testing of innovative solutions.