Frequently asked questions
Everything OEMs ask us — about data, reliability, adoption and the commercial model.
Data & security
No. Resolv is on-premise: inference, knowledge base and logs stay on your systems. No telemetry, no cloud service involved.
Yes, even in a fully isolated (air-gap) setup. It is designed for production lines where connectivity is not guaranteed — or not welcome.
On the hardware you choose, inside your perimeter: plant server, workstation or industrial panel PC. No calls to external inference services.
Every event — questions, approvals, version changes — lands in a queryable audit trail, exportable as an HMAC-signed archive: non-repudiable, ready for compliance processes.
Key-based access with separate roles: operator, approving expert, administrator. Every action in the audit is attributed to the person who performed it.
Answer reliability
Every answer cites its source — document and page — with a visible confidence score. The technician can always open the source and verify.
It answers that it does not have enough information in the knowledge base, instead of improvising. For a system used while a machine is down, an honest refusal is worth as much as a correct answer.
A measure of how well the answer is covered by the retrieved sources, shown next to each citation. Typical values are between 0.85 and 0.95: it helps the technician calibrate trust, it does not replace it.
You do, through Gate Review: every piece of content extracted from documents is approved or rejected by one of your experts before entering the knowledge base.
The knowledge base is versioned into signed atlases: you publish a new version and activate it with a hot-swap in seconds, without stopping anything. If something is off, you roll back to the previous version.
Answers are filtered by machine model: the same question on two different models draws from the right model's sources. No out-of-context answers.
Adoption & platform
No: components install with dedicated installers, you upload the manuals, you approve the content in Gate Review. Technicians start querying Resolv in days, not months.
PDFs — service manuals, technical bulletins, annotated diagrams — plus notes and procedures written directly by your technicians in the Knowledge Hub.
The interface has two gestures: typing an alarm code or writing a question. The real training is about the process — what to upload, who approves — and it is part of onboarding.
Industrial touch panel PCs, laptops and workstations, with dedicated installers for Windows, macOS and Linux.
Italian, English and Spanish, with live language switching in the interface.
Resolv works from alarm codes and questions: it does not require PLC or fieldbus integration to deliver value. Deeper integrations are evaluated within the partnership.
Commercial model
Through a partnership, not a cart: it starts with a demo on your content, the scenario is defined — internal use, embedded or both — and configured together. Every partnership includes onboarding and direct support.
It depends on the scenario, the number of technicians or machines and the content to bring on board. Pricing is defined in the proposal, after the demo: a deliberate choice, not an omission.
The interface's logo, name and theme become yours: for the end customer it is your machine's diagnostic system. It is the embedded scenario, designed for OEMs.
We load one of your manuals and let one of your technicians ask the questions on your content. Watching Resolv answer on your own manual is the most honest way to evaluate it.
Three architectural differences: Resolv runs on-premise (data never leaves), it answers only from the approved knowledge base while citing sources (no unverifiable generic knowledge), and every action is tracked in a signed audit. It is a diagnostic tool, not a general chatbot.
Yes, and it is the typical path: the service team first, with immediate ROI, then the same knowledge base goes on board the machines you sell, white-labelled.