Skip to content

Legal

AI & Data Use Policy

Last updated October 22, 2026

Future Built Digital uses AI as an engineering and business tool where it is useful, bounded, and appropriate to the work. We do not treat every workflow as an AI problem, and we do not treat model output as automatically correct.

This page explains our default position on AI use and data handling. Client-specific agreements may impose stricter requirements.

1. Where Future Built may use AI

AI may support work such as:

  • summarizing a Project Brief;
  • drafting or refining non-sensitive working material;
  • software development and code review;
  • structured extraction from documents or text;
  • classification or enrichment;
  • research assistance;
  • testing and evaluation;
  • product features specifically designed to use models;
  • internal automation and operational workflows.

Whether AI belongs in a client system is a design decision. We prefer deterministic software when deterministic logic is the safer, clearer, or more maintainable solution.

2. Project Brief summarization

The Start a Project flow may use an AI model to turn structured answers into a short prospect-facing summary.

The summarization workflow is designed so that:

  • only project information needed to create the summary is sent to the model;
  • the prospect's email address and phone number are not sent to the summarization model;
  • the model is instructed to preserve uncertainty and not invent project facts;
  • the model does not decide whether Future Built accepts or rejects the inquiry;
  • the model does not set project price, scope, ROI, urgency, architecture, or timeline;
  • the prospect can review and correct the summary before submission;
  • the form can fall back to a deterministic summary if model generation fails.

The structured answers remain separately stored from the generated summary.

3. Client and confidential data

Our default principle is minimum necessary data.

For confidential or sensitive client information, we aim to:

  • use only the data needed for the agreed task;
  • use approved project systems and service accounts rather than casually moving sensitive material into unrelated tools;
  • avoid putting client secrets, credentials, or regulated data into consumer/public AI interfaces unless the client has explicitly authorized that use and the handling is appropriate;
  • scope additional safeguards when the work involves regulated, high-risk, or unusually sensitive data;
  • respect client restrictions on model providers, hosting regions, retention, or human review when agreed in writing.

Do not send Future Built passwords, API keys, private keys, payment-card information, Social Security numbers, protected health information, or similarly sensitive data through ordinary email or the public Project Brief.

4. Model-provider training and secondary use

Future Built does not intentionally use client-confidential data or Project Brief submissions to train a Future Built general-purpose model.

When using third-party model services for business or client workflows, we prefer service configurations and contractual terms that do not use API/business inputs and outputs to train general-purpose models by default.

Provider terms and technical settings can differ by provider and product. Where model-provider data handling is material to an engagement, we will address it in the project architecture or agreement rather than making a blanket claim that applies to every provider forever.

We do not authorize AI providers to use Project Brief data for advertising on our behalf.

5. Human responsibility

AI can assist with interpretation, generation, or analysis. It does not remove Future Built's responsibility for the work we deliver.

Our engineering principle is:

Model output is untrusted input until the surrounding system handles it appropriately for the consequence.

Depending on the workflow, that may include:

  • schema validation;
  • deterministic business rules;
  • source grounding;
  • canonical-value resolution;
  • logging and traceability;
  • human review;
  • explicit failure states;
  • fallback behavior.

Not every AI feature requires every control. The required safeguards should match the risk of the action.

6. Automated decisions

Future Built does not use the public Project Brief to make fully automated decisions with legal or similarly significant effects.

We do not use the Project Brief model to:

  • automatically reject prospects;
  • infer protected characteristics;
  • create hidden personality profiles;
  • determine creditworthiness;
  • make employment decisions;
  • calculate an undisclosed lead score that decides whether a person receives a response.

If a future product introduces materially consequential automated decision-making, it should receive separate product, legal, and privacy review before launch.

7. AI in client systems

For client projects, the correct AI boundary is determined through discovery.

Relevant questions include:

  • What is the model actually being asked to do?
  • What information is sent to it?
  • Which system owns authoritative state?
  • What decisions remain human?
  • How is output validated?
  • What happens when the provider is unavailable or wrong?
  • What should be logged?
  • What evaluation is required before launch?
  • What retention and provider terms apply?

Future Built may recommend not using AI when a deterministic or existing platform solution is better.

8. Data retention and logging

AI-related logs should be proportionate to the operational need.

Where practical, logging should separate:

  • operational metadata;
  • prompt/model/version information;
  • user or client content.

Sensitive content should not be duplicated into telemetry merely because it is convenient.

Project Brief retention follows the Privacy Policy.

Client-system retention follows the applicable project architecture and agreement.

9. Intellectual property and AI-assisted development

Future Built may use AI-assisted development tools in our engineering workflow.

AI assistance does not change the project's agreed ownership model. Client-specific deliverables, Future Built background IP, reusable components, third-party software, and open-source dependencies remain governed by the project agreement and applicable licenses.

Because AI-generated material may not be exclusive in the same way as wholly original human-authored material, we do not represent that every isolated model-generated fragment is uniquely protectable intellectual property.

10. Changes and questions

AI tools and provider practices change quickly. We may update this policy as our systems or vendor choices change.

Questions about AI or data handling may be sent to:

Future Built Digital
Email: dakota@futurebuiltdigital.com