Extracted content flow
AI tools need extracted text and page structure. Users should see this before content enters an AI workflow.
AI data
AI PDF tools need explicit content handling.
AI features are not live in Phase 1. This policy page sets the product contract before summarization, chat, extraction, translation, and comparison launch.
Provider use
Content should only be sent to AI providers when the user starts an AI action.
Training and retention
Provider training and retention defaults must be documented for every AI backend before launch.
Embeddings and delete-now
Vector indexes should be user-deletable, with status feedback after deletion.
Enterprise controls
Enterprise plans may add regional processing, customer-managed keys, custom retention, and audit logs.
Backend status
AI runtime readiness
Backend readiness data shows AI provider and pgvector posture without claiming live provider execution or production vector evidence.
- AI provider executionAI provider execution is disabled or missing explicit provider policy/credentials.blocked
- pgvector runtimepgvector is not configured for production runtime execution.blocked
- Worker services0/4 worker service URLs are configured; 0 active worker heartbeats advertise 0 ready operations. 0/0 required runtime binary contract(s) are locally approved. Queue-worker HTTP dispatch is blocked by 4 missing readiness item(s). Live retained worker deployment evidence is still required.missing
- Object storageazure-blob storage signing probe generated upload/download contracts without exposing storage secrets; live write/delete/read-after-delete evidence is still required.partial
- Retained production evidence0 valid retained candidate manifests; 0 required evidence items missing.configured
Build local-build-unset; release domain api.clearpdf.net; retained evidence missing items 0.