Agent sites now display your logo in the top nav by default, with your brand name as a fallback. The new "Brand display" setting under Agent Site → Setup → Design lets you choose between logo, name, or both — so the public-facing site reads as yours rather than a generic template.
The in-app chat agent can now fetch a knowledge source by ID and read its full deep analysis: topic coverage, content structure, competitive analysis, writer guidance, and gap analysis. This was already available via the MCP tool; it's now available directly in the agent chat once ingestion completes.
The /yelp-study research page has been updated with an expanded dataset — 1,950 markets across 5 AI surfaces, now including Gemini's ungrounded surface. Gemini's per-answer Yelp-advertiser-inclusion rate closely matches ChatGPT's search-off rate, which is meaningful: it suggests Yelp's footprint is baked into model weights rather than pulled in at query time.
Several backend write paths were consolidated from per-row loops into bulk operations. Visibility scans, social engagement syncs, and topic creation all write in fewer round-trips now. The outputs are identical — scans just run faster and put less load on the database.