Site records and what's actually on site drift apart over time. Here's how AI-assisted analysis, checked by a human at every step, found around £1m in missed revenue across the first 500 sites.
Site records are supposed to reflect what is actually installed. In practice, they drift. Equipment gets added without the paperwork catching up, decommissioned kit stays on the drawing long after it has been removed, and nobody notices either way until a capacity assessment, a lease review or a structural survey forces the question. By then, the gap between record and reality has usually been sitting there for years, quietly costing money or blocking decisions that did not need to be blocked.
A Tower co needed an accurate, verifiable picture of what was physically present across their site portfolio, and a reliable way of finding where that picture diverged from the records already held.
The digital twin catalogued what was actually there. AI analysis clashed it against the paperwork.
DAS used digital twin surveys to capture every structure in detail, then used that model to identify and catalogue everything actually on site: antennas, dishes, equipment cabinets and ancillary kit, all as it genuinely exists rather than as it was last recorded.
That catalogue was then clashed against the existing drawings and inventory using AI analysis capable of ingesting multiple documents in multiple formats, rather than requiring records to be reformatted first. The output flagged the delta between what the records said and what the survey found, and every flagged discrepancy was reviewed and verified by a human before anything was updated back into the legacy systems of record.
Automating the comparison made it possible to do this at a scale that manual auditing never could, without giving up accuracy, because a human still checked every result before it changed a record. Across the first 500 sites surveyed and analysed, that process identified around £1m of missed revenue from equipment that nobody knew was generating it, and released significant structural capacity on sites where people believed equipment was installed and it was not.
The value here is not just tidier records. It is money that was sitting unclaimed and capacity that was sitting unused, both invisible until the digital twin and the AI analysis behind it made the gap visible enough to act on.
