Guardrails before automation
Loresta is designed around least privilege, tenant isolation, and auditable actions.
Social access
Connections use provider OAuth. Access and refresh tokens are encrypted with AES-GCM, scopes are recorded, and capabilities are independently gated so a connection never implies unlimited action.
Automation safety
Automated messaging is inbound-only by default. Explicit, payment, legal, safety, minor-related, and other sensitive requests return to the creator. Automatic sends remain disabled until provider review and workspace approval are both confirmed.
Useful automation with visible boundaries
Approved by you, identified where the surface supports it, and returned to you when judgment matters.
Approved by you
Creators can review and change the information Loresta may use before a draft or automatic reply relies on it.
Product evidence: Onboarding requires an explicit review before discovered public details become approved context. Creator Page chat and connected-inbox replies receive the current approved facts and offers. A reply outside the approved boundary is not eligible for automatic sending.
Limitation: AI can still misunderstand a question. When approved context is missing, conflicting, or uncertain, Loresta must avoid an automatic answer and return the conversation for review.
Available when: Use this claim only where creator approval is recorded and the approved context is supplied to the reply decision.
Automation stays visible
Supporters see an automation notice and a Loresta label when an eligible Creator Page reply is sent automatically.
Product evidence: Automatic Page chat requires a disclosure before the first eligible reply. Loresta-authored messages have a distinct sender type and visible label. Disclosure events and final delivery decisions are recorded for review.
Limitation: A label explains that Loresta assisted with the message; it does not certify the reply as accurate, safe, compliant, or creator-authored.
Available when: This public claim currently applies to Creator Page chat. Do not apply it to a connected provider unless the delivered message or provider surface can identify the automation to the recipient.
Judgment comes back to you
Loresta treats a no-send decision as a successful outcome when the next step needs human judgment.
Product evidence: Payment, legal, safety, minor-related, personal, business, and low-confidence messages are not automatically answered. Creators can pause automatic sending without deleting their setup. Needs you states keep the conversation and the reason visible for human review.
Limitation: Automated classification is not perfect. Creators remain responsible for reviewing their Inbox, setting appropriate boundaries, and pausing automation when circumstances change.
Available when: Use this claim only for workflows with a no-send outcome, a creator-visible review state, and a working pause control.
Operational evidence
Publishing, replies, policy changes, connection changes, and conversations that need the creator produce audit records and delivery receipts without copying access tokens into logs.
Workspace isolation
Every dashboard query and mutation is bound to an authenticated workspace membership. Sensitive changes require same-origin requests, CSRF validation, role checks, and rate limits. Delivery jobs use idempotency keys and conditional claims to prevent duplicate sends.
Report an issue
Send security reports to security@loresta.co. Include the affected URL and a concise reproduction without sharing private supporter data. Read Data and privacy for retention, export, and deletion behavior.