Anthropic Just Changed Marketing Forever
A B2B content agency founder maps Anthropic's recursive self-improvement loop onto marketing — and builds a three-node system that compounds output without compounding effort.
June 6thAnthropic shipped direct session-to-session messaging in Claude Code, so separate windows can hand off context, ask each other questions, and stay in sync without you copying anything by hand.
Claude Code sessions can now send messages to each other directly, turning context handoffs, context-budget management, and multi-window collaboration into things the sessions coordinate instead of things a user copies by hand.
Claude Code sessions can now message each other directly instead of requiring a human to copy-paste context between windows. The presenter demos sending a message between two sessions, then walks through three real uses: resetting a session before its 1,000,000-token context window gets too full and expensive, spinning off a new idea into its own window without losing the current thread, and keeping two sessions synced when they're both editing the same external resource, like a Google Sheet. He closes on naming every session with /rename so /resume stays browsable later.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →
Introduces cross-session messaging: one session sends data to another with no manual copy-paste, demoed by sending "hi" from session X to session Y. Frames it against a bigger use case — a master session planning a three-city trip (Rome, Paris, Lisbon) while per-city worker sessions research and stay synced on one shared budget.

Explains why the context window matters: a fuller window makes the model dumber and each message more expensive. Demos /context to check usage, warns that 40-50% full is where things degrade, and gives a concrete cost example — reopening a stale session can cost around $10 in tokens on one model tier. Distinguishes sending a short handoff note between sessions from running /compact, which re-reads the whole conversation and costs more.

Covers two more real use cases: spinning a new idea into its own session window without derailing current work, and keeping two sessions synced when both are editing the same Google Sheet so they don't overwrite each other. Mentions building his own Chrome extension now that the barrier to building has dropped. Introduces the coworker analogy — sessions as employees who knock on each other's door when they need something, then go back to independent work.

Confirms cross-session messaging needs no approval step. Recommends naming every new session immediately with /rename, then demos /resume listing sessions by name so old work is easy to find. Closes with a direct call to try the feature.
Claude Code sessions can now message each other directly, turning three constant chores — moving context, protecting your context budget, and keeping parallel work in sync — into things you describe instead of things you manually copy.
“Anthropic just solved one of the biggest problems with Cloud Code.”
“if you will send a very simple short message, you will pay $10 in tokens for this message”
“the simplest analogy for your different sessions within Cloud Code, it's like different employees within the company”
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
Separate Claude Code sessions used to be islands — if you wanted one to know what another had figured out, you copied and pasted it yourself. This video walks through Anthropic's fix: sessions that can message each other directly, and three situations where that actually matters.
An analogy for how cross-session messaging should be used day-to-day: sessions stay independent by default and only interrupt each other for a specific ask.
“Go test it out this feature for yourself. And if you enjoyed the video, here's another one prepared specifically for you.”
soft outro — no product pitch, just a try-it-yourself nudge plus a video-to-video interlink
00:02
00:11
00:16
00:20
00:27
00:33
00:39
00:45
00:51
00:57
01:03
01:09
01:15
01:21
01:27
01:33
01:41
01:45
01:50
01:56
02:02
02:08
02:14
02:19
02:25
02:31
02:35
02:43
02:49
02:54
03:00
03:06
03:12
03:18
03:23
03:29
03:35
03:41
03:47
03:50
03:58
04:04
04:10
04:15
04:21
04:27
04:33
04:38
04:44
04:50
04:56
05:02
05:07
05:13
05:19
05:25
05:30
05:36
05:42
05:48
05:53
05:59
06:05
06:11
06:17
06:21
06:29
06:35
06:41
06:47
06:53
06:59
07:05
07:11
07:17
07:23
07:29
07:35
07:43
07:46A B2B content agency founder maps Anthropic's recursive self-improvement loop onto marketing — and builds a three-node system that compounds output without compounding effort.
June 6thA five-minute walkthrough of Claude Code's Output Styles feature — the config Anthropic's own team reaches for when responses turn into a jargon-dense wall of text.
August 5thA hands-on walkthrough of Impeccable, the open-source Claude Code design skill, and its new Live Mode and Worlds features for turning AI-generated web design from generic to genuinely good.
August 4thAn Anthropic engineer quietly revealed that Claude Code deleted over 80% of its own system prompt — and shipped a built-in /doctor skill that does the same cleanup on your CLAUDE.md, skills, and settings.
July 28thA creator who pays for every major AI coding subscription ranks Codex, Claude Code, Cursor, and Devin on models, subsidies, and harness quality — then reveals how to get your employer to cover the bill.
July 24thA YouTuber walks through the exact three-skill setup — spec, build, review — that lets Claude Code or Codex keep shipping features all day without anyone typing another prompt.
July 22nd