Senior engineers met White House officials starting June 14 to reverse an export-control order that pulled Fable 5 and Mythos 5 offline June 12 (Opus 4.8 was untouched). The first commercially deployed model killed by US export controls (per Fortune), and Anthropic is mid-IPO at a reported $965 billion. Read our take.

Profound's Zero Click data shows Claude triggers web search on only 36.6% of prompts and shares just 8% of sources with ChatGPT. If you rank outside Brave's top 10, Claude cannot see you on that query, so Brave rank tracking is now a core GEO check. Read our take.

Any third-party agent can now query M365 email, calendar, meetings, and files via A2A, MCP, and REST, billed by Copilot Credits even if you already hold a Copilot seat. That decouples the org graph from the seat license, so integration cost scales with use, not headcount. Read our take.

Launched June 14 with Accenture, Bain, BCG, McKinsey, PwC, and Eliza across three tiers, backed by a $150 million commitment and a goal of 300k certified consultants by December 31. This is OpenAI's enterprise channel buildout, mirroring AWS Partner Network — landing days before the AG subpoena hit. Read our take.

LLM-ready scraping API. Point it at competitor pages to pull co-mention density and Brave rank signals for a GEO audit — exactly the diagnostics this week's Brave/Claude story rewards. Try Firecrawl.

No-code support chatbot that now trains on your closed Zendesk ticket history. If you have a backlog of resolved tickets, that is your most direct path to a grounded answer bot. Try CustomGPT.

Kit's MCP left beta June 11 with 76 read-and-write endpoints, so Claude or ChatGPT can manage your email list directly. The affiliate pays 50% for the first 12 months. Try Kit.

The mechanism behind Profound's 86.7% citation-overlap finding, plus four concrete fixes to lift your Brave rank for the queries Claude actually searches. Start here if Claude never cites you. Read it. Try Firecrawl.

CustomGPT now trains on your resolved Zendesk tickets. We compared it against Chatbase and Zendesk's own AI by team size so you can pick without a trial marathon. Read it. Try CustomGPT.

Kit MCP left beta with 76 endpoints. This is the setup order to wire an assistant into your subscriber data without breaking your automations. Best for solo creators and small lists. Read it. Try Kit.

Bugbot is now 3x faster and 22% cheaper per run, per Cursor's June 10 changelog. We mapped where it beats CodeRabbit and where CodeRabbit's depth still wins, by team and repo size. Read it.

Per Search Engine Land, Nike appeared in 71% of "best athleisure" AI prompts while New Balance appeared in 0%, despite identical Knowledge Graph entries. Co-mention density is the lever, and here is how to move it. Read it.