Generative AI has barely become part of everyday life, and the next wave of artificial intelligence is already here. Agentic AI can operate with greater autonomy, planning what to do, making decisions, and adapting in real time to achieve a goal. For businesses, that could mean more personalized services, smoother interactions, and less time spent on complex tasks. But as AI gains more freedom to act, how do we make sure people stay in control? Orange is investing in this technology with one clear principle: AI should be useful, and it should be responsible. 
 

Agentic AI: Orange is committed to responsible and sovereign AI  

The more autonomy we give AI, the more important control, transparency, and security become. Who makes the decisions? Who’s accountable? How do we protect fundamental rights and personal data? Europe introduced the AI Act in 2024 to set clear rules around these questions.

At Orange, we’re developing AI that people and businesses can trust, in line with European values. That means investing in sovereign infrastructure so data can be protected and managed in secure environments that meet regulatory requirements.

Technology alone, however, isn’t enough to guarantee digital sovereignty. Human oversight and clear accountability are built into the way we design AI systems. Because however powerful AI becomes, its decisions need to remain explainable, traceable, and under control.


The 3 pillars of trusted AI at Orange
 

  • Responsibility, with human oversight, transparency about how AI is used, and a clear ethical framework.

     

  • Sovereignty, with reliable infrastructure and secure data.
     

  • Usefulness, with practical solutions designed around the needs of customers. 
     

 

Discover our 360° innovation culture

Research and innovation: Orange steps up agentic AI  

By 2028, Gartner estimates that one-third of enterprise software applications will include agentic AI. At Orange, we’re already getting ready for that shift.
Working through open innovation, we’re teaming up with partners to develop AI solutions that are more powerful, more secure, and easier to use.

With Mistral AI, we’re working on the network infrastructure needed to process richer data and support services that respond instantly. We’re also using OpenAI models as we explore ways to deploy sovereign AI that’s efficient and less resource-intensive. Because this approach can be adapted to different needs, it can also support digital inclusion through models built around local realities, including multilingual AI solutions in Africa.

Innovation is happening inside Orange too. Since 2023, our teams have been experimenting with AI through Live Intelligence, our secure AI platform. Developed in partnership with LangChain, the new Live Intelligence Studio offering helps organizations take the next step into agentic AI.

More is coming in 2026. Colibri and Swish.ai, two start-ups selected at Orange OpenTech in 2025, will soon deploy their solutions within the Group. Their technology will help us monitor AI agents more effectively and improve IT support and customer service.

 

 

Responsible AI in practice  

Ethics and responsibility shape the way we innovate with AI. Every project follows our Data and AI Ethics Charter and is assessed by our Data and AI Ethics Council.

 

« The adoption of our Data and AI Ethics Charter reflects the Orange Group’s commitment to developing innovative technologies that serve our customers within a responsible, ethical framework and protect the trust we’ve built with all our stakeholders. »

Christel Heydemann
CEO of Orange


 

Training and initiatives such as AI BoostCamp are helping us build a shared approach to AI that’s innovative, secure, responsible, and firmly under human control.

Orange employees are leading the way
 
65 000

employees trained in AI and data as of June 2026

2 000

data and AI experts

How agentic AI is changing services and networks

Agentic AI could change the way telecom services and networks work, making networks smarter, customer interactions smoother, and day-to-day operations more efficient.

 

“Agentic AI represents a significant evolution, enabling systems to act, reason, and adapt.”

Diego Olaya
Offer Strategy Lead GenAI, Orange Business

Use case 1: more reliable, responsive networks 
 

Imagine a mobile network outage or sudden congestion. Now imagine a network that can optimize itself and spot potential problems before they happen. 

That’s no longer science fiction. Agentic AI can use predictive maintenance to anticipate outages, detect risks, plan corrective action, and adjust network settings. If an incident does happen, it can respond in real time, track the recovery under human supervision, and even keep customers informed. 

The result: incident resolution times could be cut by 20% to 50%, with problems detected and fixed faster and better service for customers. 

 

Use case 2: smoother, always-on customer service
 

Imagine customer service that can solve a problem at any time of day or night.

Generative AI typically responds to a request. Agentic AI can go several steps further. It can analyze a complex situation, access customer data, work out what’s causing the problem, and either resolve it immediately or suggest the right solution. 

Through text or voice bots, it can handle a large share of customer requests and pass more complex cases to a human advisor when needed. 

The result: smoother, more personalized, and more effective customer service. 

 

Use case 3: AI agents at work for businesses 
 

Imagine AI agents that are always available and designed around a company’s specific needs and ways of working. 

That’s already happening. Live Intelligence allows businesses to design, deploy, and manage their own AI agents within trusted infrastructure. The agents can automate tasks and plug directly into business processes, freeing up teams to focus on work where they can add more value. 

The result: measurable improvements in reliability, efficiency, and productivity.

Glossary   

AI developed and used according to ethical principles, including transparency, human oversight, data protection, and respect for fundamental rights.

AI designed and deployed in a way that protects data, technology, and infrastructure, often at a national or regional level.

The ability of a country or organization to control its data, technologies, and digital infrastructure without relying too heavily on external providers.

An autonomous AI system that can perform tasks, interact with its environment, and use tools to achieve a specific goal.

FAQ   

Generative AI creates content, such as text, images, code, or audio, in response to a prompt. Put simply, generative AI explains; agentic AI acts. By combining multiple models, it can learn, reason, take action, carry out a series of steps, and interact with other systems.

Agentic AI is likely to play a major role in the future of AI. It can automate entire processes, improve operational efficiency, and deliver more personalized services. For telecoms, the practical opportunities range from improving network performance to transforming customer service.

Trusting autonomous AI requires clear rules around transparency, human oversight, data security, and fundamental rights. Orange’s approach to responsible AI draws on European best practice and is backed by practical measures including the AI Pact, our Data and AI Ethics Charter, and our Data and AI Ethics Council.

Agentic AI can automate certain tasks, but it still needs human oversight. It’s there to complement teams, which continue to play a key role in supervision, decision-making, and handling complex situations. The model is evolving from humans being “in the loop,” providing direct assistance, to humans being “on the loop,” maintaining oversight.

Agentic AI can help optimize networks by anticipating incidents, improve customer service with support available around the clock, and automate business tasks.