AI assists the decision. A person makes it.

AI in our products explains, summarises, drafts and surfaces. It does not act irreversibly on its own. Where an AI feature proposes something — a message, a schedule change, a finding, an explanation — a person reviews it before it has consequences.

• No AI feature takes an irreversible action without explicit human confirmation.
• Automated output is presented as a suggestion, and it is labelled as generated.
• A person can always see the information the suggestion was based on.
• A person can always override, edit or ignore the suggestion.

AI output can be wrong, incomplete, or confidently misleading. That is a property of the technology, not a bug we have quietly fixed. Our products are designed so a wrong answer is visible and cheap rather than hidden and expensive.

• AI output is not professional advice — not legal, medical, financial, tax or security advice.
• AI explanations describe what information shows; they do not predict outcomes.
• AI features vary by product, and some products use no AI at all.
• AI capability described on a product still in development is direction, not a shipped feature.

Where a product touches a regulated field — WealthKraft is the clearest example — we are explicit that no advisory, brokerage or regulatory status is claimed.

• An AI feature receives the minimum information it needs to answer the question in front of it.
• Information stays inside the product that holds it; products do not pool data for AI training across the portfolio.
• We do not sell information, and we do not use customer information for advertising profiling.
• Credentials and provider keys stay server-side and are never exposed to the browser.

Our general handling of personal information is described on the privacy page.

• No covert surveillance features, and no monitoring of a person without their knowledge.
• No inference of sensitive personal characteristics for targeting or scoring.
• No automated decision that materially affects someone without a human in the loop and a way to contest it.
• No claim that an AI feature is more capable, more accurate or more certified than it is.

SFGOC Labs holds no AI certification, no independent model audit, and no compliance attestation for its AI features. This page states our practices and commitments. It is not a certification, and it is not legal advice.

Questions or concerns about an AI feature: email billing.labs@sfgoc.org, or use the contact routes.