• Considerable efficiencies mean huge time savings on repetitive tasks, thanks to the near-instant generation of reliable and accurate reports for real-life interventions.
In recent years, Orange and AMA have been working together to equip Field Engineers with innovative solutions designed to facilitate their work.
The solution uses some of the most successful LLMs on the market to quickly generate a relevant and representative intervention report.
Lightening the load
The collaboration began with remote assistance via the AMA XpertEye tool, using smart glasses equipped with augmented reality cameras. More recently, it has progressed with blockchain-based applications dedicated to notarising jobs or delegating responsibility. Generative AI now takes its place in this technological toolbox, focusing on a peripheral—but nevertheless essential—aspect of the profession: intervention reports. “Field Engineers work on jobs back-to-back. They only get time for this task at the ends of their rounds”, explained Philippe Delbary, Director in charge of the “Future of Work” programme at Orange. “At the end of the day, they may be tired, their memories hazy and their reports therefore incomplete, fragmented or inaccurate. In order to avoid these problems and lighten the engineers’ loads, AMA builds on the data captured during operations using generative AI. A reliable and faithful account of how the intervention actually took place is produced quickly and automatically”. This report comes pre-drafted: AI is used to save the engineers time, but they control the text with the ability to modify and edit it at their leisure.
Large language models in action
When Field Engineers use AMA XpertEye to collaborate with colleagues or experts, a lot of data is generated via audio and video streams. This information, contextualised in advance, is the raw material that generative AI ingests, analyses and synthesises. In this case, the solution relies on some of the most advanced LLMs (Large Language Models) on the market — after integrating GPT-4o, AMA also announced compatibility with Mistral’s open-source LLM. From a dedicated graphical interface, users will be able to access all the recorded data and obtain a relevant and representative report of their work, via a that can be adapted to meet their needs. The result is almost instantaneous, saving Field Engineers precious time on a repetitive, low-value task.
A valuable knowledge-management tool
Aside from automated report generation, the tool—which will be presented at SIDO 2024—can also be used as a search engine. “XpertEye’s video conferencing component is used to collect relevant content and data”, explained Guillaume Campion, Vice President of Global Sales at AMA. “What we seek to gain with generative AI, in addition to the reporting functionality, is a knowledge-management tool that enables us to make the most of this volume of data, in the form of a purpose-built search engine. Field Engineers will be able to search this knowledge base of all past interventions via a chatbot, enabling them to access advice and locate procedures or precise information. Once more, the aim is to make life easier for Field Engineers by providing them with a kind of User Manual 2.0. More broadly, this tool will facilitate onboarding for new starters and enrich the training systems of corporate customers”.
The solution is therefore likely to be valuable in many business sectors. The current use case may only concern Field Engineers, but other maintenance professionals also stand to gain from it, as well as audit and inspection professionals. All these roles systematically involve writing reports, which is often perceived by the workers as a tedious and repetitive task. Generative AI can facilitate and accelerate content production, while often improving quality.
A prompt refers to the written command or instruction given to a generative AI system to shape its response.
AMA
