What is a Virtual Product Twin?
A live model of one of your machines, built from your own engineering files. It connects the drawings, programs, models and documentation into a single source your teams can ask questions of, rather than hunting across separate files.
What do you need from us to build one?
The engineering files you already hold for a machine, such as drawings, automation programs, parts lists and manuals. We build the model from those. Nothing needs to be rewritten or re-documented by hand.
Can Horizon IR create all my documentation?
Yes. Once the twin is in place, the Document Engine generates drawings, manuals and spares lists from the model, and keeps them in step as the machine changes, so you can stop maintaining documents like spares books by hand.
Our documentation isn’t great — can we still use this?
Yes. The model is built from the underlying engineering files and first principles. Our AI can reason about what aspects of a drawing might be missing and suggest corrections to make the model complete. In practice the twin can become the reliable source your documentation never was.
Is our data kept private?
Your files and the model built from them stay yours. We treat every machine’s data as confidential to your organisation and do not share it between customers. All LLMs used are private to us and your important IP doesn’t actually leave your network — we extract what is needed for the model and leave the source with you.
Which team is it for?
Aftersales, manufacturing, engineering and sales. Each team asks the model in their own terms, from a fault on site to a build sequence at the bench to a technical answer during a quote. In practice, starting with Aftersales — which tends to be mostly a consumer of information — is easiest.
We don’t have an ERP/FSM/CRM package — can we still use this?
Yes. The twin works on its own and does not depend on you having an ERP, FSM or CRM system in place. Where you do run those systems, it can connect to them, but none of them are a requirement to get started.
Why don’t we build this ourselves?
The hard part is not a chat interface, it is turning raw engineering files into a reliable, connected model of a machine. LLMs are good at working with text, but very little of your products are described as text. Doing it in-house means not only building the capability but ensuring the safety of your IP and ongoing maintenance.
Why can’t ChatGPT or Claude do this?
General AI models have no knowledge of your specific machines. Ask them about your equipment and they guess from public information, which is often wrong and impossible to trust on the floor. The Virtual Product Twin grounds the answers in your own engineering files, so responses are specific to the machine in front of you rather than a plausible-sounding average.