A Frugal Approach to AI: Nomadia's Eco-Responsible Choice

25/06/2026
11:49

reading time : 2 min

Lucie Monnot

Content Marketing Manager

Artificial intelligence is making its way into everything, at the cost of ever-growing consumption of hardware resources and energy. Discover how Nomadia limits the environmental impact of its solutions and increases their operational efficiency through a pragmatic and considered frugal approach to AI.

AI is not just LLMs!

Since 2022 and the arrival of ChatGPT on the scene, mainstream media discourse has reduced artificial intelligence to large language models (LLMs) and generative AI applications. This conveniently overlooks everything that came before: 40 years of progress in computer vision, speech-to-text and text-to-speech, analytics, and machine learning, built on proven algorithms that are far less energy-intensive than solutions based on today’s LLMs.

The ease of access to LLMs, combined with their versatility, makes it easy to forget the heavy physical infrastructure that underpins them — infrastructure that, wherever it is located, consumes an ever-growing share of available electricity, potentially at the expense of other essential uses.

"Green" AI is a mirage

Knowing that the giants of tech and AI are investing directly in electricity production — including nuclear technologies — to power their super datacenters with “green” energy only confirms their phenomenal ambitions for development and hegemony. They are all betting on an AI that is ever more powerful and, therefore, ever more hungry for energy and other resources: chips, servers, cables, satellites, datacenter ventilation and cooling systems, not to mention the construction materials for the datacenters themselves.

In fact, LLM models are becoming increasingly powerful and, as a result, increasingly attractive and less financially costly. The more a technology advances and spreads, the less it costs, which encourages ever greater use — at the price of an unsustainable environmental impact that the major players claim to offset by talking about “Green AI.”

Yet there is no viable “green” AI in a sector whose cumulative CO2 emissions are already growing at 6 to 8% per year. And when Jensen Huang, the CEO of Nvidia, announces that today’s computing power will need to be multiplied by 100 in the coming years to meet society’s AI needs, any socially and environmentally responsible company cannot help but ask: should we put AI into everything? For Nomadia, the answer is clearly no.

Nomadia's philosophy: doing more with less

For a software publisher like Nomadia, systematically turning to LLMs — regardless of the problem to be solved or the need to be addressed — amounts, in many cases, to using a sledgehammer to crack a nut, when a simple fly swatter and a minimum of skill would suffice.

“This is why we advocate for a frugal and considered approach to AI, in line with our responsibilities as a mission-driven company. It consists, whenever possible, of doing more with less, always taking into account the limits of a system that is claimed to be infinite, when it clearly is not. Eco-responsibility is a core value of Nomadia. It is not a value for show. It guides and determines every decision we make in the design, development, and use of our products. This implies radical choices, particularly with regard to generative AI.”

Stéphane Delbecque, Chief Product & AI Officer, Nomadia

The pillars of Nomadia's frugal AI

A considered and frugal use of AI necessarily begins with a questioning of use cases, which requires the effort of inventorying and prioritizing them according to their usefulness and added value for Nomadia’s solution users. Identifying a need does not predetermine the technical approach to addressing it. The question of “how” only comes in a second step, and involves testing different approaches while respecting the following principles:

Not limiting itself to LLMs, but always considering possible alternatives — typically algorithms that Nomadia can train autonomously for very specific purposes.
Using only open-source technologies, so as to maintain control over tools, processes, and data, both during testing and in production.
Working locally wherever possible, with technologies deployed on Nomadia’s own servers, in France and not in distant datacenters, for reasons of data sovereignty and customer data security.

Applying these principles leads Nomadia to systematically favor the most energy-efficient artificial intelligence technologies:

“For us, the right technology — the optimal solution — is the one that addresses each identified use case in the most frugal way possible. And when you look at things from that angle, it is not necessarily generative AI. For example, in the field of image recognition, we achieved better results by using the right algorithm and training our own model than by using an LLM, and this at an energy cost 100 times lower. That is what I call doing more with less.”

Stéphane Delbecque, Chief Product & AI Officer, Nomadia

That said, Nomadia does not deprive itself of the possibilities offered by LLMs and generative AI. True to its principles, the group has chosen to work with the French company Mistral which, in addition to its nationality, has the advantage of offering an open-source LLM perfectly suited to Nomadia’s needs and requirements. This solution is hosted in France, at Mistral, and is used by Nomadia as needed to test and run the models that enrich the features of its solutions and boost their performance.

Frequently Asked Questions

FAQ – The Most Frequently Asked Questions About Nomadia

Why Choose Nomadia?

As France’s leading publisher of Smart Mobility SaaS solutions, Nomadia supports more than 175,000 field professionals every day. Our solutions are easy to use, quick to deploy, and deliver significant and immediate return on investment.

Drawing on the expertise of both a software publisher-integrator and a consulting firm, Nomadia’s teams provide tailored support, from data consulting to the deployment of mobile devices. Finally, our technical support team is available 24/7 to assist you.

Who Are Nomadia Solutions Designed For?

Nomadia solutions support the digital transformation of all mobile professionals: field sales representatives, delivery drivers, technicians, auditors, healthcare workers, inspectors, service providers, security patrol officers, experts, and more.

Whether for SMEs or large enterprises, our solutions adapt to businesses of all sizes and across all industries.

How Much Does It Cost?

Nomadia Delivery offers transparent and flexible pricing, which primarily depends on the number of users (for example, planners and dispatchers) rather than a fixed cost per parcel.

Thanks to this model, you can control your delivery costs based on the number of resources (users) involved in route management, providing great flexibility to adjust your subscription according to the size of your fleet and your operational needs.

Is It Compatible with My Current ERP System?

Yes! Nomadia Delivery has been designed to integrate quickly and easily into your existing environment thanks to our powerful and secure APIs. Our solution is compatible with most ERP systems on the market. API integration enables the automatic synchronization of your transport orders, customer orders, and customer information, ensuring smooth and error-free delivery tracking. The solution also offers user-friendly import capabilities with data validation controls, as well as export options in multiple customizable formats.

Is It Suitable for Our Delivery Volume?

Yes, Nomadia Delivery is ideal for companies that manage a high volume of deliveries and want to optimize their routes, create balanced territories for their drivers, and ensure precise tracking of every parcel. Nomadia Delivery adapts to fleets of all sizes!

Can Multiple Warehouses or Depots Be Managed?

Yes, our solution allows you to manage multiple warehouses or depots. It centralizes data and optimizes routes or service operations for each of them.

Is There a Limit to the Number of Stops in Route Optimizations?

No, there is no strict limit to the number of stops in route optimizations. Our solution can handle large volumes of stops and quickly calculate optimized routes.

Read our latest articles: