Pondero AI logo

Pondero AI

Archives
Log in
June 9, 2026

Pondero Brief: 2026-06-09: Siri rebuilt on Gemini, Claude now an iOS default

Pondero Brief - JUNE 9TH, 2026

The $1B/year Gemini deal, iOS 27 Extensions, and Anthropic's $965B IPO filing.
pondero. BRIEF
Siri rebuilt on Gemini, iOS 27 AI Extensions showing Claude as default option
AI

WWDC 2026 Put Claude, ChatGPT, and Gemini on iOS. Here Is Which Memory System Wins.

Apple's Extensions framework just gave all three assistants a system-level slot on every iPhone. The memory comparison you need before you pick your default.

JUNE 9TH, 2026 · BY JONATHAN HILDEBRANDT

WWDC 2026 dropped two bombs. Siri now runs on a custom 1.2-trillion-parameter Google Gemini model, under a deal MacRumors pegs at roughly $1 billion per year. But the bigger move for this audience: iOS 27's new Extensions framework lets users pick Claude, ChatGPT, Gemini, or Grok as the default AI for Apple Intelligence features (Writing Tools, Image Playground, and Siri itself), per AI Weekly. Claude on over a billion Apple devices, as a system-level option, for the first time. Developer betas went live June 8.

Privacy routing splits three ways: simple tasks stay on-device, moderate requests go to Apple Private Cloud Compute, heavy reasoning hits Google Cloud. Queries are processed statelessly and Google cannot train on the data, per The Next Web.

Why it matters. ChatGPT, Claude, and Gemini now compete at the OS level on every iPhone. Our comparison of how each one handles memory across sessions is the natural next read.

See the full memory breakdown →
 

Anthropic files for IPO at a $965 billion valuation.

JUNE 9TH · PONDERO NEWSDESK

Confidential S-1 hit the SEC on June 1. The Series H raised $65 billion at a $965 billion post-money valuation; Fortune reports annualized revenue approaching $50 billion. Per CNBC, Claude Code crossed $1 billion ARR within six months, and over 1,000 customers now spend more than $1 million per year. Bankers are targeting a fall 2026 debut above $1 trillion. If you're evaluating Claude for enterprise work, the window before IPO pricing pressure is narrowing.

 

Opus 4.7 matches specialist chemistry software on NMR prediction.

JUNE 9TH · PONDERO NEWSDESK

Anthropic's latest research paper shows Opus 4.7 matching MestReNova (the standard tool in organic chemistry labs) on carbon NMR and beating it on hydrogen chemical-shift assignments. Zero-shot, no fine-tuning. On sub-peak spacing, Claude landed within half a hertz 80% of the time; ChemDraw and MestReNova managed 26-35%. A general-purpose LLM encroaching on specialist lab software is a real signal for biotech procurement teams weighing Claude Enterprise.

 

OpenAI and the Trump White House negotiate a government equity stake.

JUNE 9TH · PONDERO NEWSDESK

Talks center on a proposed Public Wealth Fund that would give the government a stake in OpenAI and distribute AI returns to citizens (The Decoder). No terms finalized. Separately, Senator Sanders is drafting a 50% one-time share tax on major AI companies. For anyone comparing OpenAI vs. Anthropic as a vendor: governance just moved from background reading to a due-diligence line item.

 

Sakana AI opens a Recursive Self-Improvement Lab in Tokyo.

JUNE 9TH · PONDERO NEWSDESK

Six active research projects, including the AI Scientist (fully automated paper writing) and a Darwin Godel Machine for open-ended self-improvement (Sakana AI). Outside DeepMind and Anthropic alignment work, RSI research has been mostly theoretical. A funded startup treating it as engineering, not philosophy, shifts the timeline assumptions.

 

Make.

JUNE 9TH · PONDERO TOOLS

Route webhooks from Firecrawl /monitor, n8n, or Claude events into structured automation workflows. Our Firecrawl + Make guide this week walks through the exact build. Try Make.

 

Firecrawl.

JUNE 9TH · PONDERO TOOLS

LLM-ready scraping API with a /monitor endpoint for diff-based page watching. Get a webhook only when content actually changes. No more cron polling. Try Firecrawl.

 
ChatGPT vs Claude vs Gemini memory comparison

Which AI assistant actually remembers you across weeks?

JUNE 5TH · PONDERO COMPARISON

ChatGPT leads on passive recall with Dreaming V3 (free users included). Claude wins on inspectable, editable memory through Projects. Gemini ties to your Google account, most caveats. iOS 27 Extensions just put all three on equal footing at the OS level. Read it.

 
Cursor SDK v3.7 custom tools vs MCP servers decision guide

Cursor SDK v3.7 custom tools vs. MCP servers.

JUNE 6TH · PONDERO GUIDE

When to build inline with Cursor's SDK, when to spin up an MCP server, when to hand the whole thing to Make or n8n. The decision tree and code for each path. Read it.

 
Firecrawl /monitor with n8n or Make workflow guide

Firecrawl /monitor with n8n or Make: skip cron-polling.

JUNE 6TH · PONDERO GUIDE

Send a webhook only when a page changes. Token math and a working build for each platform included. Read it. Try Firecrawl.

 
7 tactics to cut AI coding bills in 2026

7 tactics to cut AI coding bills in 2026.

JUNE 6TH · PONDERO GUIDE

Prompt caching, batch jobs, model routing, spend caps. Per-token math for each. Read it.

 
TrustFall one-click RCE across coding agents via MCP

TrustFall: one-click RCE across every major coding agent.

JUNE 6TH · PONDERO REVIEW

A trust-this-folder click can trigger remote code execution in Claude Code, Cursor, Gemini CLI, and Copilot via MCP. The mechanism, why Anthropic calls it working-as-designed, and the lockdown settings worth checking now. Read it.

 
Jonathan Hildebrandt Jonathan Hildebrandt
Co-founder and primary operator of Pondero. Writes the Pondero Brief.

Affiliate disclosure  ·  Unsubscribe  ·  Manage preferences

Pondero earns commissions on some links. This does not affect our editorial picks.

Don't miss what's next. Subscribe to Pondero AI:
pondero.ai
Bluesky
LinkedIn
Twitter
LinkedIn
Powered by Buttondown, the easiest way to start and grow your newsletter.