Beat: industry deltas, last 24–48h only (labs/people/hardware/capital/policy). Model & platform releases = Dispatch's; robotics depth = Sol's.

Aproape nimic cu adevărat nou azi pe beat-ul ăsta. I swept all five angles (open net, frontier figures by name, hardware/edge, capital, policy). Everything that surfaced as "fresh" is one of three things: already on the board, Dispatch's lane, or an article from this week about an event from last week. No new lab/people move, no new chip, no new round, no new bill in the 24–48h window. So today is mostly the skip-list — but two threads are worth a sentence because they're crossing thresholds, not because they broke today.

The one thing brewing that's worth your radar

• Agentic payments crossed the "any merchant" line. Visa embedded its rails directly into ChatGPT (~Jun 10) so an agent can pay at any Visa-accepting merchant — no merchant opt-in, which is exactly what killed OpenAI's earlier Instant Checkout. winbuzzer Jun 11
• The read (and why it's brewing, not news): this is the latest beat of a year-old drumroll — Mastercard Agent Pay (Apr '25), Visa Trusted Agent Protocol (Sep '25), Google AP2 (Sep '25, 60+ partners). You almost certainly know the rail exists. What's new in kind is the threshold: bypassing merchant adoption means the bottleneck moves from "will merchants integrate" to "who issues the agent's consent record + spending guardrails." The moat is becoming the permission layer, not the payment. Second-order for us: that's the exact design we already live inside — Claude Code's permission modes, spending limits, approval-before-action. The whole economy is converging on the guardrail pattern we treat as plumbing. Underlying event is ~72h old → strictly past my 48h cutoff, flagged as brewing not as a delta.

Quietest-interesting structural thread

• Government equity stakes in frontier labs (Public Wealth Fund concept; Trump confirmed talks Jun 5–6, OpenAI named). Week-old, so not a delta — but it's the structurally weirdest thing in the air and it's still un-resolved. The angle worth holding: if Washington takes a direct stake in OpenAI, it re-prices every frontier-lab IPO including Anthropic's — a sovereign cap-table line item is a different risk object than a normal float. Watch whether this firms up before Anthropic's window. TechCrunch Jun 6 · WaPo Jun 5

Already known / not my beat / too old — skipped

• SpaceX SPCX trading (Jun 12) — on the board from 06-12.
• Google Personal Intelligence + AI Mode expansion across Gemini/Chrome (Jun 12–13) — platform release = Dispatch's beat.
• Microsoft MAI in-house models (MAI-Thinking-1 etc.) — Jun 3, and a model release → not mine, not new.
• Trump gov-equity-stake / Public Wealth Fund — Jun 5–6, surfaced above as brewing only.
• Fei-Fei Li / World Labs $1.23B + Marble — Feb/Jun 3, old.
• Mistral / SSI / TML — no June 12–13 move; an aggregator's "Mistral raised $100M four weeks after inception" is garbled (wrong founder name, wrong founding year) → discarded as unreliable, not reported.

For us specifically

• Nothing touches the stack hard today. The agentic-payments permission-layer convergence is the only item with a real "this is our design pattern going mainstream" read — worth a line in any EthBuilds/Dispatch positioning thinking, not an action.