Frequently asked questions
about .expert.

They are wearable devices with a camera, microphones, audio and a HUD display designed for use at industrial plants. The operator works hands-free while a remote expert sees exactly what they see and overlays instructions, diagrams and annotations in real time.

Yes, .expert works with smart glasses designed for industry. Resistance to impacts, dust and liquids, and ATEX certification for potentially explosive atmospheres, depend on the chosen model. The model is selected according to the scenario: workshop, plant maintenance, railway environments, buildings or environmental plants.

Battery life depends on the chosen model. For long shifts, external or swappable batteries can be used without interrupting the session (hot-swap), again depending on the chosen model.

Yes, they are used hands-free. Compatibility with helmets, ear defenders, masks and prescription glasses depends on the chosen model. Interaction is via voice commands and physical buttons, without a touchscreen.

It means that the language models (LLMs), the vector database and the RAG pipeline are installed and run on physical or virtual servers inside the customer's network. Documents, conversations and images stay within the company perimeter.

We mainly use open-weights models (for example from the Llama, Mistral and Qwen families), run on-premise and connected to the customer's documents via RAG. On request we also evaluate proprietary LLMs, provided they can run locally.

Answers are grounded in company documents through a RAG (Retrieval-Augmented Generation) pipeline: every answer cites its source (manual, data sheet, procedure) and the LLM is configured to say "I don't know" when the context is insufficient. No source, no answer.

It depends on the number of users, the volume of documents and the workloads: typically one or more servers with enterprise-class GPUs. Larger organisations use redundant infrastructure. Sizing is defined after an initial analysis.

Yes, through API integration with ERP, MES and CMMS, assessed case by case. For example, the AI can read work orders and bills of materials and record intervention notes in the systems already in use.

No. The entire pipeline (LLM, vector database, content archive, logs) runs on-premise: models and data stay on the customer's servers. Live sessions between the operator and the remote expert can be routed over the customer's VPN or dedicated networks.

The platform is designed to process data on the customer's servers. Privacy roles and data processing agreements are defined contractually with each customer, together with the privacy notice for operators being filmed and the tools to handle data subjects' rights. Video recordings have configurable retention periods.

Only users authorised through roles and permissions. Recordings are optional and can be disabled per shift or per area; rules for access logs and retention are defined during the project.

Models and data stay on the customer's servers; encryption, access logs and compliance requirements (e.g. NIS2, ISO 27001) are defined during the project together with the customer's IT team (for example, segregation between OT and IT networks).

The return depends on downtime avoided, site visits saved and faster training: we estimate it together during the pilot project.

We work with an annual fee that includes software licences, AI model updates, technical support and Customer Success. The cost depends on the number of users, sites and modules activated. Hardware (smart glasses and servers) can be purchased or rented.

Mean time to repair (MTTR), number of interventions resolved remotely, training hours saved, time to bring a new operator up to speed, knowledge base usage and documentation coverage of the plant. KPIs are defined together and made available in dashboards or exportable to BI tools.

The duration is agreed together based on the scope, typically a few weeks: document collection, guided use on a real scenario (for example a plant or a maintenance team) and review of the results with the roll-out plan.

Yes. We organise a demo based on your real processes, on-site or remotely: uploading a manual, a live session with smart glasses and natural-language questions to the knowledge base.