Release 1.8.0 · Media intelligence replaces legacy analysis; org chat moves to a chat service; landing pages gain deeper routing

This release replaces the older media-analysis path (Video Intelligence + BigQuery–style usage) with a media intelligence stack built around multimodal embeddings (Gemini Embedding 2), pgvector-backed artifact storage, modality-aware usage, and updated accrual/billing cron jobs. Repository and brand flows that touched “media analysis” now align with the new model, including billing constants, usage schemas, and UI copy where surfaced.

Org AI chat is wired toward a dedicated backend chat service entrypoint plus frontend configuration and documentation so streaming chat can run as its own deployable surface. The dashboard org chat route and model picker/settings continue to evolve with the new service contract.

Landing page authoring receives a broad set of builder, library, routing, experiments, and settings updates, including a catch-all routing segment under the landing-page detail tab so LP URLs can carry richer routing tails without losing the shell. Treat landing page development as still experimental: behavior, URLs, and guardrails may change as the product hardens.

Control plane work improves directive evaluation publishing, signal/execution API shapes, and schedule-related typing; a database migration adjusts control-plane directive scheduling (next_run / dropped schedule fields). Review migration notes before deploy and ensure workers/cron pick up the new handlers.

Across the dashboard you may notice translation and navigation tweaks tied to these areas; admin and org surfaces were touched where experimental routes, usage, or integrations intersect.