Define the boundary
Choose the routine questions the assistant may handle and the conversations that always return to a person.
Integration Planned
Give AI agents a narrow creator DM capability instead of unrestricted inbox control
Loresta is planning an API for authorized agents. It will expose controlled inbox actions while keeping provider credentials, policies, and tenant data behind Loresta's boundary.
An agent should receive a capability, not the keys to everything
General AI agents are useful orchestrators, but unrestricted access to private social inboxes creates unnecessary privacy and account risk. The integration layer should scope every credential to a workspace, creator profile, and explicit action set.
Loresta's planned API will keep provider access behind its own authorization layer and expose revocable credentials with rate limits and audit records.
- Workspace scope
- Creator profile scope
- Action scope
- Revocation and rate limits
Structured outputs make downstream automation safer
Instead of returning only free-form text, a creator DM API can return intent, confidence, proposed action, draft, handoff reason, and an optional approved media identifier.
External agents can then make deterministic decisions, such as notifying a manager when a collaboration arrives or requesting approval before a draft is sent.
- Intent and confidence
- Draft and action
- Handoff reason
- Approved media identifier
Public documentation must define the no-send cases
A credible API reference needs more than endpoint syntax. It should describe which topics cannot auto-send, how retries work, what counts toward usage, and how message history is retained or removed.
Those contracts are part of the planned launch. Private application routes are not a substitute for a supported public API.
- Safety contract
- Idempotency contract
- Usage contract
- Retention contract
A controlled rollout
Use the smallest safe workflow first, then widen it only after real conversation review.
Add the creator context
Approve the voice, facts, destination, and optional media that can support a useful reply.
Connect carefully
Authorize the supported inbox with official access and begin in draft-first mode.
Review and improve
Inspect replies, handoffs, usage, and repeated intent before widening any automation rule.
Launch checklist
Every connection and automatic reply should remain authorized, observable, and reversible.
- Official account authorization
- Inbound-only messaging
- Creator-approved facts
- Editable voice profile
- Conversation context
- Low-confidence handoff
- Personal and business inquiry handoff
- Visible usage and pause controls
Frequently asked questions
Direct answers about capability, rollout status, and creator control.
Can an AI agent use Loresta today?
Not through a supported public API yet. The agent API is planned and private dashboard endpoints must not be treated as public integrations.
What will the API return?
The planned design includes structured intent, confidence, action, draft, handoff reason, and optional approved media references.
Will agents receive Instagram or X tokens?
The intended architecture keeps provider credentials inside Loresta and exposes only scoped Loresta capabilities.
How will agent actions be controlled?
Through scoped credentials, role and workspace checks, rate limits, idempotency, audit logs, and product safety policies.
Prepare the workflow while the integration is planned
Loresta keeps voice, facts, media, policies, and human handoffs visible before automation expands.