FAQ
Questions asked first
Short answers to what enterprises and labs usually ask us. For anything else, write to us.
What does NEV Research Labs do?
We are a commercial rights syndication and AI research firm. Today we acquire rights to enterprise data and IP, and use them in our own research and with our partners. Next, we map how each company works — who does what and how information moves — so buyers get an understood company rather than a data dump. Where we are headed is simulation labs: de-identified digital twins of real operations, built to each country's data law, where models train on tasks that are genuinely hard.
What is operational IP?
The record of how a company actually runs: its data, its workflows and the architecture of the systems that connect them. Calls, tickets, transactions, drawings, approvals — and, crucially, the outcomes attached to them. It is valuable because of its volume, complexity, the number of tools involved and the processes it records, and most companies do not yet know the rights to it have value.
We are an enterprise. Do we have to sell anything?
No. The data stays yours. We acquire the rights to use it for agreed purposes, such as training AI, and may use those rights ourselves or license them to downstream partners. We never take custody, never compete with you, and keep everything confidential on the terms we agree with you. Everything is de-identified to your local law before it moves, and every term is agreed in writing with your counsel.
What is a company blueprint?
An understood company. Instead of raw records, a buyer gets what the company does and who it serves, the roles and personas inside it, how work and data move between people, and annotated samples of real workflows. It is the bridge between rights-cleared data and a full simulation.
What is a simulation lab, in plain terms?
A de-identified digital twin of a real business that an agent can act inside. It has personas, tool clones, hard tasks drawn from real work, and a verifier that checks the result against what actually happened. The agent tries, gets scored, and tries again. It is where we are headed, not something we run today.
Can you help us train AI on our own work?
Yes. Alongside data licensing, we are starting to work with companies directly: we build evaluation environments from your own archives, run current models against them to see where they fall short, and help you train agents on your own work, locally. This runs independently of licensing, and can combine with it if you want both.
Do you train models or run compute?
No. We build the environments and the facility around them. Labs bring their own models and compute, train, and leave.
Where does the data come from, and what rights do you hold?
From operating companies, from businesses that are winding down or changing hands, and from IP holders. Rights are acquired outright or under a licence that explicitly covers derivative datasets and environments, because de-identification alone does not grant the right to build on data. Ownership, copyright and licence terms are confirmed before anything moves.