Featured Story
What the $500B Stargate AI plan could mean for telecom
The flood of investment into AI data centers powered by Nvidia chips could have major implications for 5G silicon developers.
Orange Business is adding a new solution to its GenAI portfolio, targeting smaller companies with ready-made use cases, multiple LLMs and secure data handling.
With generative AI adoption growing, telcos seem to be getting involved with the technology in myriad ways. Orange Business has recently introduced a multi-LLM GenAI solution offering secure handling of data and a database of pre-defined prompts, targeting smaller businesses.
Launching early next year in France and then throughout the first quarter in the rest of Europe, Orange Business is targeting "hundreds of customers" with the new multi-LLM SaaS solution, Aliette Mousnier-Lompré, the CEO of Orange Business, said in a press conference during the operator's recent OpenTech event in Paris. It also forms part of what has newly been labelled the Live Intelligence range, which will encompass all of the company's GenAI offers.
The basis for the product was Orange's internal use of AI. The company has been making multiple LLMs available to its employees in a secure environment since last year. It now counts 50,000 internal users, she said, of whom 12,000 are active weekly. Users are told they can access generative AI in a responsible way, with the cost and CO2 impact of each query communicated to them.
It is aimed at smaller businesses which, Mousnier-Lompré said, are finding the GenAI landscape difficult to navigate. To make it accessible, the SaaS solution is fully managed with a library of pre-defined prompts, which range from basic to sophisticated use cases.
For example, the tool may analyze an email thread and write a summary, or access legal expertise, scanning an NDA, say, to highlight areas that may need special attention. Other uses include writing a job description and creating a cross-analysis of supplier responses to an RFP.
Having a library of pre-defined use cases will be one of Live Intelligence's differentiating features, according to Mousnier-Lompré, who argued that it will help companies generate more sophisticated outcomes than if they accessed GenAI models directly.
Orange Business says it guarantees to customers that their data or intellectual property will be safe from leaks. "With our solutions, [customers] have the guarantee that no company data, no company intellectual property, will be used to feed the AI models," Mousnier-Lompré said.
The solution is intended to give customers flexibility regarding the model used. It will offer customers a choice between numerous GenAI models, allowing them to choose based on their budget and needs. They include Mistral, ChatGPT and Gemini, and more may be added in future.
Assessing impact
At the moment, the package merely includes hosting, the LLM and the user interface, but in future it may be bundled with connectivity or even APIs, according to Mousnier-Lompré. Speaking about pricing, she said that while Live Intelligence Trust comes at fixed tariffs, the new SaaS solution will have a more dynamic pricing model based on usage and the selected models.
Orange's CTIO, Bruno Zerbib added that all-you-can-eat pricing models are ill-suited to GenAI, which comes with the constraint of GPU capacity: "It creates a major challenge where we have to re-educate all of our users of artificial intelligence to make sure that they understand exactly the complexity of their use case with regards to how much GPU they're going to need."
Orange is also very cautious about the environmental impact, Zerbib noted, adding emissions will be an important factor for the company when working with AI companies going forward.
The customers targeted by the offer are mainly companies with "specific requirements in terms of trust," according to Mousnier-Lompré, with the public sector, health sector, utilities, energy, defense, banks and insurance amid the main targeted verticals.
This is not a new focus for Orange Business, which in March launched an infrastructure-as-a-service (IaaS) solution which is now called Live Intelligence Trust, aimed at French businesses that want stronger protection around their data. This product, which was launched together with French start-up LightOn, ensures customers retain complete control over a trusted infrastructure, dedicating GPUs to the solution in Orange's cloud platform. In the second half, the infrastructure-as-a-service already generated "several millions of euros in business," according to Mousnier-Lompré.
Offering generative AI to businesses isn't an entirely novel proposition for a telco. Vodafone, for example, started offering Microsoft's GenAI technology to enterprise customers, while Deutsche Telekom offers a service called Business GPT for security-conscious customers. However, it seems neither include the same breadth of models.
Orange Business reshaping
The Live Intelligence offer comes as Orange Business shifts its strategy, with an overhaul initiated when Mousnier-Lompré took over in 2022 to reshape the unit and respond to a decline in legacy business. In the third quarter of 2024, Orange Business reported a 2.6% drop in revenues, and the unit seeks to halve its EBITDAaL losses this year before returning to growth in 2025.
The new strategy is for Orange Business to act as an integrator of networks and digital solutions, focusing on end-to-end services spanning connectivity, as well as "digital" areas like cloud, cybersecurity and AI, offered on demand via APIs and platforms.
Network and infrastructure are increasingly converging with tech and data components for customers, according to Mousnier-Lompré, who demonstrated the point through the example of a hypothetical customer with a large factory who wants to implement real-time computer vision.
They may want to install video cameras and AI engines to detect anomalies in real time, possible requiring very low latency for security reasons, she said, adding it would also require placing AI at the edge close to the plant.
Orange Business seeks to offer companies solutions addressing both sides of the equation, such as data handling, potential use of cloud and latency requirements.
You May Also Like