AI Data Disclosure

1. Lead disclosure

When you choose an AI-assisted feature, Precision Foundry may send relevant project information, instructions, and existing work content to OpenAI so the requested suggestion can be generated. Do not include sensitive personal information, regulated data, credentials, private keys, or confidential information unless your organization has determined that doing so is appropriate.

Effective date: 2026-08-20.

2. Which features use AI

AI-assisted features in Precision Foundry may include, depending on your plan and workflow:

• Project story and value builders and analyzers
• Business and IT estimate helpers
• Initial / portfolio readiness review analysis
• Portfolio and requirement discovery questions, refinement, and conversation helpers
• Stakeholder and delivery-team identification helpers
• Requirements generation and clarification
• Delivery planning, epic/story/criteria/task generation, and duplicate analysis
• Issue analysis and issue discovery helpers
• Seed-context clarification helpers

Some workflows include heuristic fallbacks that can operate without calling OpenAI when the AI provider is unavailable or a call fails. Heuristic output is still a draft and requires human review.

3. What information may be sent to an AI provider

Supported AI features use the OpenAI API endpoint chat.completions.create. Requests may include system instructions and a user message that contains structured project, issue, discovery, requirement, or delivery context needed for the feature—such as titles, descriptions, prior answers, requirements text, or related artifacts already in the workspace.

Avoid unnecessary regulated, highly sensitive, or secret information in fields that may be included in AI requests.

4. OpenAI API use and training

Precision Foundry uses the OpenAI API for AI-assisted functionality.

OpenAI states that information submitted through its business API is not used to train or improve its models by default unless the API customer explicitly opts in. Precision Foundry does not opt in to model training through its application code.

Precision Foundry sends supported API requests with store: false. In application code, every chat completion request forces store: false so prompts and completions are not retained by OpenAI for distillation, evals, or dashboard logs through that store setting.

Setting store: false does not necessarily eliminate provider abuse-monitoring logs. OpenAI may retain abuse-monitoring information for up to 30 days by default, unless legally required to retain it longer or different approved controls apply.

This disclosure does not say that OpenAI immediately deletes all data, and it does not claim Zero Data Retention unless a separate OpenAI account configuration is confirmed outside this repository.

5. What Precision Foundry stores locally

Distinguish these categories carefully:

• Information sent to OpenAI: feature-specific context and instructions needed to generate a suggestion.
• Generated content returned by OpenAI: draft text or structured suggestions returned to the application.
• Content saved in the Precision Foundry database: drafts and analyses your workspace keeps as normal project/delivery records after generation.
• AI execution metadata and cost records: operation labels, model, token counts, and cost amounts used for allowances and overage billing. Full prompts and outputs are intentionally omitted from the AI execution ledger.
• Heuristic features: local draft helpers that may run without OpenAI.

6. How generated output is handled

When you accept or save AI-assisted results in a workflow, those results become ordinary workspace content subject to the Terms of Service and Privacy Policy. Until saved according to the feature's workflow, temporary UI state may exist only in the session.

7. Human-review expectations

AI output may contain errors or unsupported assumptions. Precision Foundry does not automatically approve projects, commit funding, or make final business or technical decisions. A person must review and approve generated content before it becomes authoritative.

AI-generated content is a draft. You remain responsible for reviewing, editing, validating, and approving it before relying on it.

8. Cost tracking

Precision Foundry tracks AI usage in dollars based on actual OpenAI cost. Plans include an allowance; usage above the allowance is billed at actual OpenAI cost plus 25%. Workspace billing views show included, used, remaining, and overage information where available.

9. How to ask AI-data questions

Privacy and AI-data questions: privacy@precisionfoundry.io

General support: support@precisionfoundry.io

Also see the Privacy Policy (/privacy) and Terms of Service (/terms).