Not a plausible-sounding model, real title records
A lot of "AI property intelligence" claims are really just a plausible-sounding model with no real data behind a specific claim. Ownership is a good example of where that shows: knowing who actually owns a site, and whether it’s a UK-registered company, isn’t something you can infer, you either have the real data or you don’t.
Lumeyrie ingests HM Land Registry’s full corporate-ownership dataset (CCOD), 4.4 million title records, and checks every site against it directly. If there’s no match, the report says so; it doesn’t guess.
The seven evidence domains
Power & grid connection
Checks whether sufficient grid capacity and a credible connection path are plausible within the required time and cost.
Planning history
Checks local planning evidence, relevant site history, brownfield or allocation signals, and where policy evidence is still missing.
Environmental constraints
Screens flood zones, Green Belt, SSSI, ancient woodland, historic landfill and other mapped environmental constraints.
Fibre & connectivity
Checks coverage, network proximity and route-diversity signals to test whether adequate fibre looks deliverable.
Transport access
Measures access to major roads, junctions, rail and airports so logistics constraints are visible early.
Ownership & tenure
Checks corporate ownership against HM Land Registry records and identifies when a title-specific search is still required.
Noise sensitivity
Screens nearby sensitive receptors and likely acoustic constraints before a specialist BS 4142 assessment.
Every domain in a Lumeyrie report is scored and evidence-tiered on its own, then rolled up into one recommendation, not averaged into a single number that hides where the evidence is thin. For the full explanation of how confidence tiers work and why the hard-stop layer exists, see Trust & Evidence.
Have a listing you’re not sure is worth the deeper work? Send it over, the first one’s on us.
Send a listing