Hope everyone has a great week! The GTM Engineer Lab is hosting its first AMA this week on Thursday at 11am EST with Eric Nowoslawski on 📝cold email. Don’t forget to apply to the lab if you’re interested in joining! I’ll be doing a round of acceptances on Weds, so you can make it in time for the AMA.

Favorite resources from the week:

• Jacob Dietle’s step by step starter guide to setting up your Context OS
• Nate Herkleman summarized Andrej Karpathy’s notes to 10x Claude Code effectiveness1️⃣ Create a folder with two subfolders: raw (your source material) and wiki (where the LLM organizes everything)2️⃣ Drop in source documents, articles, transcripts, whatever you want3️⃣ Tell Claude Code to ingest the raw files and it builds out wiki pages with relationships, tags, and backlinks automatically
• James McAulay shared a Reddit screenshot on why Opus 4.6 might be worse than it was a few weeks agoTl;dr: the Anthropic plans were running at a huge loss, so they put in lower quality models
• Casey Hill on the power of tables and screenshots in Lovable’s FAQsLoveable dropped in tables + screenshots of pricing, which is a better user experience and easy for LLMs to parse
• Alex Lieberman’s simple, powerful Claude Skill he runs every MondayScan Linear, Notion, Slack, Gmail, and Cowork sessions to identify repeatable processes that should become new Claude skills
• How Understory advertises incentives while avoiding scammersThey’ve identified the specific words that scammers look for in LinkedIn ad libraries (like taking: “reward” and “gift card” out of the text)
• Srivatsan Venkatesan on the importance of Google shipping WebMCP in Chrome 146A new way for websites to tell LLMs exactly how to parse them. Companies that implement WebMCP will have an advantage while everyone else catches up
• I had Michel Lieben onto the podcast to talk about LinkedIn content, build vs. buy, tech stacks, how ColdIQ is using Claude + Cursor, and more