AI usage
Agents use AI to generate structured strategic proposals scoped to their domain and your 90-day goals. No agent output is published or executed without founder review and explicit approval.
AI Trust Center
Agents propose. Founders decide. Everything is logged.
AgentCompany is designed around a bounded-autonomy model. AI agents generate strategic proposals on a 90-day planning cycle, but no agent takes action on spending, publishing, or product changes without explicit founder approval. Supported proposals, decisions, and outcomes are designed to produce an audit trail that belongs to you.
Data handling
Goals, signals, proposals, decisions, and outcome records are stored in Supabase behind server-side routes. Agent outputs never include raw credentials or system secrets.
Payments
Subscriptions are managed through Stripe. AgentCompany stores access state and subscription tier — not raw payment credentials or card data.
Goals and signals
The 90-day goals and domain signals you define as inputs for agent proposals. Stored per cycle and per app.
Agent proposals
Structured proposals generated by domain agents, including rationale, recommended actions, and expected outcomes. Stored in full for audit.
Founder decisions
Every approval, partial approval, or rejection logged with the founder's annotation, a timestamp, and the proposal it references.
Outcome records
Cycle-close notes and outcome data logged against approved proposals. Used to inform the next cycle's goal-setting.
Account records
Login identity, subscription access, connected DVGProOS apps, and support context.
Billing records
Stripe customer IDs, subscription state, invoice status, and webhook event references.
Legal and responsible-use notice
Human accountability stays in the workflow
Policies control. This page is a transparency summary. The Privacy Policy and Terms of Service govern use of AgentCompany™.
AI requires approval. Agent proposals may be inaccurate or incomplete and are not professional advice. An authorized human must review consequential actions.
Rights and authority. Connect only systems, content, and data you own or are authorized to use.
Security and vendors. Safeguards reduce risk but cannot guarantee absolute security. Approved providers may process task data needed to perform requested operations.
Your responsibility. You remain responsible for privacy, employment, advertising, intellectual-property, consumer-protection, and industry obligations.
Names and marks. AgentCompany™ and DVGProOS™ are claimed trademarks. Third-party marks belong to their respective owners.
Proposal audit trail
Keep a durable record of proposals, decisions, and outcomes.
The proposal audit trail is designed to preserve agent proposals, founder decisions, annotations, and cycle outcomes. Availability, export, retention, and correction behavior depends on the product version, account state, Privacy Policy, Terms of Service, and applicable legal requirements.