AI prepares the work. Product Foundry reviews the risk.
FoundryAI provides decision support across the marketplace. It helps organize evidence, explain recommendations, and identify next steps; it does not replace user confirmation, talent consent, or Product Foundry review.
Last updated: August 12, 2026
Draft notice
These plain-language beta terms describe the current Product Foundry workflow. They remain subject to applicable agreements and should be reviewed by qualified Canadian legal counsel before being treated as final legal terms.
Company Copilot and brief intelligence
- • Structures pasted, uploaded, conversational, and form-based company intake into the existing brief workflow.
- • Records source, confidence, confirmation state, missing fields, uncertainty, contradictions, and version history.
- • Suggests focused clarification questions and may recommend a pricing model without changing the company's selection.
- • Does not invent or silently overwrite confirmed budgets, dates, locations, skills, deliverables, or requirements.
Talent Copilot and Talent Pro
- • Assesses profile completeness, readiness, evidence gaps, preferences, availability, and recommended next actions.
- • Labels recommendations and requires talent acceptance or editing before profile information changes.
- • Records rejected or deferred recommendations so they do not silently become profile facts.
- • Does not invent or verify employers, titles, dates, credentials, skills, accomplishments, rates, availability, work authorization, references, or proof points.
- • Talent Pro tools remain optional and entitlement-gated.
Explainable matching
FoundryAI may combine deterministic rules with consent-controlled semantic signals to assess multiple fit dimensions and prepare evidence, confidence, positive signals, concerns, possible exclusions, missing information, and audience-specific explanations. Confirmed hard exclusions require confirmed evidence and review. When semantic data is unavailable or disabled, deterministic matching remains available.
Shortlists and introductions
Raw AI recommendations cannot independently progress a company-facing introduction. Product Foundry reviews recommendations before an approved ShortlistItem becomes visible. Personalized opportunity explanations are reviewable and cannot fabricate talent interest or company promises. Talent must approve any expression of interest, and contact details remain withheld until consent and required approval are complete.
Ask Foundry
Ask Foundry answers from concise, current information the authenticated user is authorized to access. It filters private and unapproved data, limits input and context, treats retrieved text as untrusted, applies rate and timeout controls, and falls back to deterministic current-status answers when AI is disabled or unavailable. It cannot complete a workflow action merely by claiming that it did.
Feedback, calibration, and outcomes
Structured feedback may propose calibration signals for future recommendations. A single event does not silently become a permanent rule. Signals remain bounded, inspectable, approval-controlled, reversible, and traceable when used. Hard requirements and confirmed exclusions remain separate from learned preferences.
AI audit and administrator review
Material AI activity may record task type, provider, model, prompt version, input or output references, confidence, status, errors, retries, review state, reviewer, and override reason. Product Foundry limits secrets and unnecessary personal data in logs. Authorized administrators can review, accept, reject, or override material outputs according to their role.
Human decisions and prohibited uses
AI does not make final hiring, rejection, compensation, credential-verification, introduction, contact-release, contract, payment, fee-waiver, or administrator decisions. Product Foundry does not use protected or sensitive personal characteristics as matching or ranking signals.
Limitations and user review
AI output can be inaccurate, incomplete, biased, stale, or unsuitable for a particular purpose. Scores and explanations are recommendations, not guarantees of qualification, availability, legal compliance, or engagement success. Companies, talent, and administrators should correct errors and rely on confirmed workflow status and applicable agreements.
Questions and choices
Users can review this disclosure, correct their own information through supported workflows, and contact Product Foundry about AI processing. Optional AI features and consent-controlled semantic processing may be disabled or withdrawn where the platform provides that control, although doing so may reduce related functionality.