CDBGet an estimate

For AI labs and model builders

UK construction records with a clean rights chain.

Licensed datasets from UK main contractors and specialists, covering how real projects were tendered, programmed, varied and delivered. Every record is cleared for rights and anonymised before it reaches you.

What you get

Data you can train on and defend.

Real site records, not synthetic

Data from UK contractors doing the work: how jobs were tendered, programmed, varied, inspected and handed over, including the mistakes.

Linked at the project level

Tenders and BoQs, programmes, RFIs and technical queries, variations and compensation events, site diaries, site imagery, CVRs and final accounts from the same projects, joined by consistent IDs.

Anonymised by default

People, clients, addresses, planning references, title blocks, photo location data and faces are removed with one documented method, following ICO anonymisation guidance. Project facts that could pinpoint a building are banded.

A licence that covers training

UK copyright law only permits text and data mining for non-commercial research, so commercial training needs a licence. Every source is licensed under a signed agreement with the contractor, with its rights basis recorded per record.

Packaged to spec

JSON or Parquet for structured records, original PDFs and images with redaction applied, programmes as XER, PP or XML, and IFC models where the rights allow. Schemas and labelling matched to your training or eval pipeline.

Ready for evals and environments

Linked outcomes make natural tasks: price a BoQ and compare with the final account, answer a technical query, forecast a programme slip, assess a compensation event.

Licensing

Start small, then scale.

Evaluation
Anonymised samples under NDA before you commit.
Non-exclusive
Standard licence, lowest price, fastest to close.
Exclusive
Sole rights to a dataset for an agreed term or permanently.
Custom collection
Tell us the trade, region or record type you need and we'll source it.

Request data

Tell us what you need.

We reply within one working day with available supply and a sample plan.

Record types, trades or sectors, regions, volume, and whether you need exclusivity.

Samples are anonymised and covered by an NDA.