Issue #29

August 24, 20264 min read

Anthropic eyes $100B IPO

Claude Code now resumes sessions automatically when your usage limit resets, and new Concise output style cuts narration while keeping the same results. Anthropic is preparing for a blockbuster IPO, enterprise growth is outpacing OpenAI on coding workloads, and Gmail and Drive connectors are expanding.

Anthropic Could Aim to Raise $100 Billion in Blockbuster I.P.O. - The New York Times — New York Times

On this page

Highlight
Worth Knowing
Under the Hood
From Anthropic
In the News

Highlight

Sessions resume automatically when your usage limit resets

If you hit a claude.ai usage limit mid-task, Claude Code now picks the session back up on its own once the limit resets, instead of leaving the turn stranded until you return. You can turn this off under "Continue automatically at usage limit" in /config. The usage-limit message also now tells you when your session or weekly limit resets, not just the monthly one.

Worth Knowing

New Concise output style cuts narration

A built-in "Concise" style makes Claude lead with results and skip the preamble and running commentary while doing the same work. You can select it under Output style in /config.

New environment variable sets your default starting model

ANTHROPIC_DEFAULT_MODEL sets the model new sessions begin on without locking you in the way ANTHROPIC_MODEL does; a /model pick still overrides it and persists across restarts.

More slash commands now open mid-turn

You can open /permissions and /add-dir while Claude is working, and permission rule changes apply to the rest of the current turn. In the fullscreen TUI, /autocompact, /theme, /help, /config and /advisor also open mid-turn, and their panels no longer cover the latest messages.

Readline keybinding flavor matches Bash word behavior

Setting keybindingFlavor to "readline" makes Ctrl+W delete back to the previous whitespace as in Bash, and aligns Alt+F, Alt+D, Ctrl+Y and word-wise arrow movement with shell conventions. The default "classic" behavior is unchanged.

VS Code transcript adds screen reader support

The VS Code extension now announces replies, permission requests, errors, and status changes to screen readers, and supports per-turn heading navigation through the transcript.

Under the Hood

Memory growth in long sessions capped, claude-api skill down to 25k tokens

Long interactive sessions no longer grow without bound because subagent tool results are released once they leave the display window, expired WebFetch content is dropped as intended, and the built-in claude-api skill now costs roughly 25k tokens to load instead of 200k+.

Caching, billing, and startup fixes

Prompt caching works again behind an LLM gateway or custom base URL, and a Bedrock streaming bug that silently doubled billed API calls behind Content-Type-stripping proxies is fixed. Startup is faster on macOS, with the automatic update check deferred about 10 seconds so it no longer competes for CPU at launch.

From Anthropic

Claude Blog · 20 min read

The Claude Code guide for startups

Five working practices small teams use to get more out of Claude Code, from context setup to review workflows. Useful if you are trying to formalize how your team uses the tool rather than improvising per project.

Claude Blog · 20 min read

The Claude Code guide for startups

Claude Blog · 8 min read

Claude on call: How Claude Tag serves as Anthropic’s first responder for CI/CD failures

An Anthropic CI engineer walks through the agent that triages build and test failures internally. It is a concrete look at wiring Claude into an existing pipeline, including what the agent is allowed to do on its own.

Claude Blog · 8 min read

Claude on call: How Claude Tag serves as Anthropic’s first responder for CI/CD failures

In the News

New York Times · 5 min read

Anthropic Could Aim to Raise $100 Billion in Blockbuster I.P.O. - The New York Times

Anthropic is reportedly preparing a public listing that could rival the largest on record. The company's funding structure shapes how aggressively it can invest in products like Claude Code.

New York Times · 5 min read

Anthropic Could Aim to Raise $100 Billion in Blockbuster I.P.O. - The New York Times

Anthropic is reportedly preparing a public listing that could rival the largest on record. The company's funding structure shapes how aggressively it can invest in products like Claude Code.

Wall Street Journal · 5 min read

Exclusive | OpenAI’s Second-Quarter Sales Show Tepid Growth Compared With Anthropic

Reported figures put Anthropic ahead of OpenAI on enterprise growth this quarter, driven largely by coding and agent workloads. Relevant context if you are making a case for Claude Code inside your organization.

Wall Street Journal · 5 min read

Exclusive | OpenAI’s Second-Quarter Sales Show Tepid Growth Compared With Anthropic

The Verge · 4 min read

Anthropic explains how Claude’s invisible text watermarks will work

A breakdown of the SynthID-based text watermarking Anthropic plans to apply to Claude output. Worth reading if you ship generated text or documentation into downstream systems.

The Verge · 4 min read

Anthropic explains how Claude’s invisible text watermarks will work

A breakdown of the SynthID-based text watermarking Anthropic plans to apply to Claude output. Worth reading if you ship generated text or documentation into downstream systems.

The Verge · 3 min read

Claude can do more with Gmail and Google Drive, and access to Claude Cowork is expanding.

Expanded Gmail and Drive connectors plus broader Claude Cowork availability. Useful background on where Anthropic is taking non-terminal Claude surfaces.

The Verge · 3 min read

Claude can do more with Gmail and Google Drive, and access to Claude Cowork is expanding.

Expanded Gmail and Drive connectors plus broader Claude Cowork availability. Useful background on where Anthropic is taking non-terminal Claude surfaces.

A heavy fix week across Remote Control, fullscreen rendering, and cross-session messaging, with the usage-limit continuation being the change most people will notice first.