The 3P Data Control Check

A procurement checklist for construction data and AI

7 min read · Research checked 20 August 2026

Introduction

Part I introduced the 3Ps of Data Control: Permission, Promise and Prevention. Part II showed how they apply to real construction technology products. This final part turns the framework into a practical checklist for evaluating software before sharing project information.

The checklist should be applied to a specific feature, deployment and data flow, rather than to a supplier as a whole. A provider may have permission to reuse one type of data, promise to restrict the processing of another, and offer a customer-controlled environment for a third. The questions are therefore cumulative: the aim is to understand what happens to each type of information, who can use it, and what control the customer retains.

This checklist is intended to support procurement decisions. It is not a substitute for legal, cybersecurity or professional procurement advice.

Download the checklist (PDF)6 pages · Free · No sign-up

Chart 1: The 3P Data Control Check

ModelWhat it meansQuestions to ask
PermissionThe supplier may use customer data beyond providing the service.What data is covered?For what purposes?For how long?Does this include anonymous or derived data?Did we have the right to grant that use?
PromiseThe supplier processes the data, but its use is restricted by contract and controls.Who receives the data?Are subprocessors named?What are they allowed to do with it?How long is it retained?What happens when processors or models change?
PreventionThe customer controls whether project data leaves its environment.Who controls administrators and keys?Where does the AI actually run?Can the system work without an external connection?When can data leave?Does the customer deliberately approve that handoff?

* The same product may sit in different positions for different data or workflows.

** “Vendor does not process” only applies when the customer controls who can access the data and when it can leave. A private environment run by the vendor is still Promise, not Prevention.

1. Permission

Permission covers any right to use customer information beyond returning the customer’s own result. That may include training, product development, benchmarking, research, analytics or commercial data products.

The first task is to establish exactly what has been authorised, including rights over anonymous and derived data.

Rights and authorised uses

  • What licence does the customer grant to the supplier?
  • Is that licence limited to providing the service to this customer?
  • Can identifiable or raw project content be used to improve a model, algorithm, evaluation set or product?
  • Can anonymous, de-identified, aggregated or derived information be reused?
  • Can the supplier create benchmarks, market reports, cost databases or other commercial data products?
  • Does any reuse occur only inside this customer’s environment, or across customers?
  • Is consent opt-in, opt-out, inherited from standard terms or negotiated in the order form?
  • What survives withdrawal, termination or deletion?
  • Which agreement controls if the order form, DPA, AI notice and online terms conflict?

Authority to grant permission

A customer may be willing to permit reuse without having the right to do so for every element in the file.

  • Who owns or controls the original files and the intellectual property within them?
  • Does the uploader have authority to grant the proposed rights on behalf of the client, designers, subcontractors and other rights holders?
  • Do appointments, licences or project contracts limit downstream use of the material?
  • If the supplier asks for consent to training or broader reuse, who inside the organisation is authorised to give it?

Anonymous and derived data

Do not accept “anonymised” as a complete answer. Ask what has been removed, what remains, and what new data is created from the original.

  • Which project and personal identifiers are removed?
  • Are geometry, dimensions, room relationships, structural labels or specification details retained?
  • Are quantities, prices, rates, productivity assumptions or commercial qualifications retained?
  • Are user corrections, accepted or rejected outputs, labels or embeddings retained?
  • Can an internal identifier still connect the derivative to the source file?
  • Can multiple supposedly anonymous records be combined to identify a project or reveal a repeatable design or construction method?
  • Who owns the resulting derived dataset, and can it be used across customers, licensed, sold or published?
  • Does deleting the source also delete anonymous and derived copies?

Anonymising the project does not necessarily anonymise the knowledge contained in it.

2. Promise

Promise applies when project data enters infrastructure controlled by the supplier or by providers it selects, while contracts, policies and controls restrict what those organisations may do with it.

The customer needs evidence that the supplier, and the providers it relies on, can keep that promise.

Contract and governance

  • Do the executed contract and order documents expressly restrict training, product improvement and other secondary use?
  • Does the definition of customer data cover prompts, outputs, embeddings, labels, corrections and intermediate artefacts as well as source files?
  • Do the same restrictions apply to subprocessors and AI providers?
  • Does the supplier remain responsible for the providers it selects?
  • Will the customer receive advance notice of material processor, model or feature changes?
  • Can the customer object, disable the affected feature or terminate after a material change?
  • What are the return, export and deletion obligations at termination?
  • What incident-notification obligations apply?

Trace the data

For the named feature, map the full path from source file to result. Training is only one possible use; data may still move through several systems even when training is prohibited.

  • Which hosting, storage, OCR, search, vector, model, evaluation, logging and support providers can receive project information?
  • What does each provider receive: full files, crops, extracted text, embeddings, prompts, outputs or metadata?
  • Where is each stage processed and stored?
  • What are the default and maximum retention periods?
  • Under what conditions can supplier or provider personnel access the information?
  • Can abuse monitoring, support or incident investigation trigger additional storage or human review?
  • How does deletion propagate to active storage, indexes, caches, logs, backups and derived datasets?
  • Can the relevant processor, model or provider change without customer notice?

Chart 2: Trace the Data

For customer-controlled boundary, follow one file through a feature

  1. Project fileDrawing / model / RFI / estimate
  2. Construction platformWhere the customer uploads or opens it
  3. ProcessingOCR / extraction / search / AI preparation
  4. External servicesAI model / storage / other processors
  5. Result & retentionOutput / logs / history / derived data

Questions to ask at each stage

  1. Who receives it?
  2. What do they receive?
  3. How long do they keep it?
  4. What are they allowed to do with it?
Not every product uses every stage. The point is to trace the data flow, not to assume a fixed architecture.

For each box, establish:

  • The operating organisation
  • The region of storage and processing
  • Whether raw content, a derivative or metadata is received
  • Default and maximum retention
  • Human-access conditions
  • Whether the information can be used for training, evaluation or product improvement
  • Whether the component can change without notice
  • How deletion reaches backups, indexes, logs and derived datasets

Training is one possible side effect of this journey. Exposure to another system exists even when training is prohibited.

Assurance evidence

Security evidence should be credited for the question it answers, and no more.

  • Is the relevant product and AI feature inside the scope of the current SOC 2 Type II report?
  • What is the exact organisational and technical scope of the ISO/IEC 27001 certification?
  • What recent penetration-test coverage and remediation evidence is available?
  • Which assurance reports cover material subprocessors?
  • How is privileged internal access approved, logged, reviewed and revoked?
  • Which controls depend on customer configuration or administration?

Security protects data from unauthorised use. It does not limit authorised use.

Chart 3: What Common Assurance Signals Do—And Do Not—Establish

An assurance should be credited for the question it answers and no other

Assurance signalUseful evidence ofWhat it does not prove
SOC 2 Type II reportDesigned controls operating over a stated period, subject to the report’s scope and exceptionsThat customer content cannot be used for training; that every product or subprocessor is in scope
ISO/IEC 27001 certificateAn information-security management system and risk-management process within the certified scopeA particular data-use restriction, zero retention, or customer-controlled infrastructure
Penetration testResistance of tested systems to selected technical attacks at a point in timeContractual confidentiality, processor behaviour, or the security of untested services
DPARoles and obligations for regulated personal dataProtection of every form of commercially sensitive non-personal construction information
NDA/confidentiality clauseA duty to restrict use and disclosure of confidential informationThat systems technically prevent access, or that anonymous derivatives remain confidential
Data residency commitmentWhere specified information is stored or processedOwnership, no-training, local-only support, or identical residency for logs, backups and AI calls
Subprocessor listThe named organisations performing stated functions at the date shownThat the list is complete for project content, or that changes require customer approval
No-training statementA restriction on model development, if the subject and scope are clearNo retention, no human review, no inference processing, no derived analytics, or no additional recipients
Zero-data-retention configurationThe relevant processor should not persist submitted content beyond transient inferenceThat the construction vendor, its index, its logs or its support systems retain nothing
Customer-managed keysGreater customer control over decryption and revocation, depending on implementationThat plaintext is never processed, or that the vendor has no privileged path while the service is running
Single tenancyReduced risk of accidental cross-tenant access and stronger isolationCustomer control of administrators, keys, backups or outbound processing
Local/open-weight inferenceThe model can operate without sending prompts to a hosted foundation-model APIThat the runtime has no telemetry, remote tools, update channel or other outbound connection

3. Prevention

Prevention applies when the architecture allows project information to remain inside a customer-controlled boundary rather than relying only on a supplier promise.

A “private cloud” or single-tenant label is not enough on its own. The practical question is who controls access and when data can leave.

Customer-controlled boundary

  • Does project content remain on a customer device, network, server or cloud account?
  • Who controls the root administrator and identity provider?
  • Who holds and can revoke encryption keys?
  • Can vendor personnel obtain standing or emergency access, and is that access visible to the customer?
  • Who controls outbound network rules and DNS?
  • Can the core workload operate with external network access blocked?
  • Where do licence checks, crash reports, telemetry and update requests go?
  • Can diagnostics expose file names, paths, document fragments or screenshots?

AI inside or outside the boundary

  • Where do AI inference, embeddings and conversation history run?
  • Does the model runtime call external models, tools, safety services or telemetry endpoints?
  • Can external AI be enabled only by an administrator or deliberate user action?
  • Can the user see what content is being sent and to which provider?
  • Can external AI be removed or disabled while the core product remains useful?

Prevention is about making the boundary visible and enforceable: where the AI runs, when data leaves, and who can access it.

A one-page answer every supplier should be able to provide

A buyer should not need to reconstruct the data path from multiple policies and security pages. Ask the supplier to complete the following statement for the specific feature and deployment being considered:

For named feature and deployment, the customer provides data classes.

The original is stored in systems and regions for period.

The following derived artefacts are created: list.

They are retained for period and may be used for purposes.

The following organisations receive project information: names and roles.

No recipient may use it for training / improvement / benchmarking, except clearly stated permissions. Human access is possible only for reasons under controls.

Provider changes are governed by notice / approval mechanism.

Deletion removes systems and derived artefacts within period, subject to backup / legal exceptions.

The customer can operate without this external processing by local / customer-managed / non-AI mode, or “not available”.

If the supplier cannot complete that statement clearly, the buyer does not yet understand the data boundary well enough to approve sensitive project information.

The five-minute pre-upload check

Before a drawing, model, estimate, programme, contract or project history enters a new service, answer five questions:

  • Permission — Can identifiable, anonymous or derived information be used for anything beyond this customer’s result?
  • Authority — Do we have the right to grant those uses for everything contained in the file?
  • Promise — Which organisations and systems will receive the information, and what prevents each of them from using it for another purpose?
  • Prevention — Can the information remain inside a customer-controlled environment, and if it leaves, is that handoff visible and deliberate?
  • Persistence and change — What survives deletion, and what happens if the supplier changes a model, processor, feature or term?

If any answer is “we assume”, stop and resolve it before uploading the data.

The practical principle

The 3Ps are not three labels to assign to a supplier. They are three questions to ask of every important data flow: what have we permitted, what has been promised, and what can the architecture prevent?

Either bring the model to the data, or make the data’s departure explicit.

Take the checklist with you

The full 3P Data Control Check, the five-minute pre-upload check and the supplier statement template, in one printable PDF.

Download the checklist (PDF)6 pages · Free · No sign-up

About the research

The profiles are based on first-party terms, privacy policies, security pages, product documentation and AI notices reviewed on 20 August 2026. They are a public-document audit, not a penetration test or legal opinion.

Let’s talk!

Questions about the research, or about how your product handles customer data? We’d like to hear from you.