AI Across Multiple CDEs: Why One Vendor's AI Only Sees Part of the Picture

Every major CDE vendor now offers AI features, and each of them can only see the data held in its own platform. For an organisation that runs projects across several CDEs, no single vendor's AI can answer questions about the whole portfolio.

A simple example

Picture an organisation running 105 projects across five CDEs. Its leadership asks which projects are trending over budget and what is driving the overrun. The AI in each platform can answer for the projects it holds, which might be a quarter of the portfolio at best. Nobody gets a complete answer, and the partial answers use different terms, fields and structures, so they cannot simply be added together.

Why the data is the obstacle

AI in construction is only as useful as the record underneath it, and many AI pilots stall because the data is incomplete, inconsistent or scattered. Before any model can spot delays or trace the cause of rework, the information has to meet three conditions. It must be complete across every platform, consistent in its structure and metadata, and held somewhere the organisation owns and controls. Vendor AI tools cannot meet the first condition, because their reach stops at the edge of their own platform.

What a connected record looks like

A connected record brings files, versions and metadata from every CDE into one location, translated into a single schema. ISO 19650 provides a sensible common language for status, revision and classification. Version history is preserved so questions about what changed and when can be answered. Every item keeps a link back to its source system, so answers can be checked.

Building it with the Utopiaverse™

The Utopiaverse™ builds that record in your own Azure tenancy. CDE Sync™ feeds it from live projects during delivery, CDE Backup™ maintains copies throughout, CDE Migrate™ carries history across platform changes and CDE Archive™ adds completed projects. AI Insights then searches and questions the whole record using retrieval augmented generation and OCR, and every answer cites the files it drew on.

Keeping control of the data

Running AI over project data raises fair questions about where the data goes. AI Insights runs natively on Azure inside your own tenancy, in the data region you choose, so the record is not copied out to a third party to be indexed. Utopia Digital's products stream files between your systems without storing them, and the company holds SOC 2 Type II and Cyber Essentials.

Where to start

Most organisations start with one layer. A business with a live multi-CDE programme often starts with CDE Sync™, while one closing out projects may start with CDE Archive™. Each step adds to the same record, and AI Insights becomes more useful as more of the portfolio arrives.

Previous
Previous

How Can I Enable AI to Read SharePoint Project Documents?

Next
Next

Joint Venture CDEs: Sharing Data and Keeping the Record at Close