Born West · Internal
Top Story
The NYT profiled Medvi, a GLP-1 telehealth startup that claims $1.8B in annualized revenue using a two-person team — code by ChatGPT and Claude, ads by Midjourney, customer service by ElevenLabs. Maury flagged it, the team reacted with open mouths. Then the counter-story landed: the FDA had already issued a warning letter six weeks before the profile ran, and Medvi has no official outside valuation. The Techdirt rebuttal is worth reading alongside the original piece.
The real signal under the noise: a two-person, AI-native operation can credibly appear billion-dollar scale. The scaffolding is real even if this particular company turns out to be a mess.
What's Shipping
GitHub launches Stacked PRs
GitHub's new gh-stack extension lets you break a big feature into a chain of small, reviewable PRs — each one stacked on the last. Achal called it out as especially relevant for AI-assisted large feature builds where you want the model generating chunks, not monoliths. Private beta; join the waitlist at gh.io/stacksbeta.
Shared by Achal · April 14Adobe takes on NotebookLM with Acrobat Spaces
Adobe launched a free AI study tool inside Acrobat — upload PDFs, docs, handwritten notes, and it generates flashcards, mind maps, and a two-person podcast about your material. Kunal flagged it. The interesting bit: Adobe tested with 500 students at Harvard and Berkeley before shipping. Competition in the document-understanding space just got real.
Shared by Kunal · April 7Cowork's next chapter is already past due
Maury shared a Lenny's Podcast episode with Felix Rieseberg — the Anthropic engineer who built Claude Cowork — and noted: "The Claude Cowork roadmap is one month, at most." That was April 10. Whatever's coming has either shipped quietly or is sitting in final review. The episode covers how Cowork was itself built by Claude in 1.5 weeks, and why Rieseberg thinks the tools that win in 2026 won't be the ones that think best.
Shared by Maury · April 10From the Team
Xenia built a Claude + Figma spec pipeline
Xenia connected Claude Code to Figma via the open-source Figma Console MCP, letting Claude crawl frames and components directly and generate structured dev-handoff spec pages in minutes. She also built a custom Claude skill that restyles the output to Born West brand guidelines. Two projects tested, output described as "quite strong." Dev handoff files are attached in the original thread if you want to see the quality.
Shared by Xenia · March 18Kunal's AI challenge project dropped
Kunal shared what he'd been building for the internal AI challenge — four fire reactions, highest engagement of the month. Details are in the screen recording attached to the original message.
Shared by Kunal · April 7Vibe coding vs agentic engineering
Achal drew a line that's worth keeping: vibe coding is for prototypes, agentic engineering is for production — reviewed, tested, safe to ship. Good framing to have ready when clients or partners ask how we use AI in real work.
Shared by Achal · April 2Quick Reads
Shopify goes agentic on commerce
Shopify's CEO has made AI usage a baseline expectation across the company, and the platform is being rebuilt around shopping agents that act autonomously on behalf of buyers. The X post Achal shared has since been widely circulated as a model for top-down AI culture change.
Shared by Achal · April 10Mistral launches Forge for enterprise model training
Mistral's new Forge platform lets companies build and continuously improve AI models on their own proprietary data — going head-to-head with AWS, Azure, and GCP on one of enterprise AI's most valuable battlegrounds.
Shared by Kunal · March 17Philly courts ban AI glasses
Philadelphia courts started blocking all smart AI eyeglasses as of this week. Achal's comment: "And so it begins." The courthouse-as-leading-indicator for where regulation is heading next.
Shared by Achal · March 30