Large-scale capex project data processing & a unique decision-layer dataset for AI

Vixel is the company behind Vrex, a real-time collaborative platform for capital (capex) projects across all industries — manufacturing, energy, infrastructure, healthcare, maritime and more.

Our expertise sits in two areas. First, standards-grade large-data handling: we open and work with very large project data in line with industry standards, including models and reality-captured point clouds that other tools cannot open at scale, and bring it together so that decision-makers and stakeholders — not only technical specialists — can access and understand it. Second, a proprietary decision-layer dataset: in doing the above, we generate a unique new data source — a record of how decisions are actually made across thousands of collaborative project reviews. This combination gives us a meaningful head start on the AI journey every industry is now focused on.

We can offer this capability to research and innovation partners exploring creative, AI-driven ways of working that improve efficiency, reduce rework and miscommunication, and support more sustainable outcomes across established industry activities.

Crucially, the thousands of conversations and micro-decisions captured across collaborative reviews don't just speed up projects — they become a living operational record. During operations, that record translates into actionable insight: why an asset is configured the way it is, what was considered, and what changed over time. And because the data is standards-based and persistent, it carries through the full circular asset lifecycle — into major maintenance, decommissioning, reuse and material recovery, where understanding the as-built and as-operated reality is what makes sustainable, circular decisions possible.

This matters across two very different time horizons. For buildings and infrastructure, assets live 20+ years, so a faithful decision history bridges generations of owners and operators and informs eventual decommissioning long after the original team has moved on. For manufacturing facilities, the rhythm is far faster: major maintenance is frequent, and the demolition and construction of process lines is almost routine — so the ability to reuse decision data and reconfigure circularly, again and again, compounds in value.

Increasingly, the project model itself becomes a kind of sensor for a review: it reflects where a team's collective attention converges or scatters, which parts of the design are genuinely engaged with, and which questions are raised but never closed. We are turning that into precise, explainable risk signals — surfacing the weak signals that minutes and transcripts miss, early enough to act on them.

A simple illustration: in one facility, a second floor was being discussed, structural load metadata was opened, the structural discipline wasn't in the room, and no follow-up was logged. Individually, four faint signals; together, a red flag worth catching before it reaches site. That is the kind of outcome we want to make reliable — fewer surprises, less rework, and more sustainable, circular decisions across the asset lifecycle.