How Vercel’s Eve runs agents from markdown

#1 𝕏

Guillermo Rauch demos “eve,” a minimal agent defined entirely in markdown (an instructions.md plus a skills/your-expertise.md) and deployable with one Vercel command. He argues markdown is set to become the next hot, most accessible programming language ever.

#2 𝕏

Harrison Chase shares @jit_infinity’s Leve: a filesystem-first, durable agent framework built on LangGraph that treats each agent as a directory of files, compiles it into an executable agent, and is inspired by Vercel’s Eve.

#3 𝕏

Harrison Chase highlights a nearly 10-hour agentic AI course covering LangChain, LangGraph, RAG, deepagents and guardrails. He’s also asking for other strong Lang* resources for learners.

#4 in

Peter Yang introduces HyperFrames’ free, open-source tool that uses Codex and Claude Code to generate videos in pure HTML. He highlights key workflows—like creating a frame.md and storyboards—and teases a full tutorial tomorrow.

#5 in

Udi Menkes created CEO-Bench, a 500-day simulation that gives an AI Agent a virtual $1 M startup to autonomously set pricing, invest in marketing, improve product and infrastructure, and close enterprise deals under noisy, delayed feedback.

#6 𝕏

Jason Zhou notes that companies are currently limited to running AI agent loops individually on each laptop, but argues this short-term setup makes it essential to build new infrastructure for a true “multi-player agent experience.”

#7 𝕏

Madhu Guru highlights an identity crisis in product—old-school PMs use AI to pump out more PRDs, strategy decks and docs with little added judgment, while Builder PMs deploy AI agents for market/user research, analytics and ideation across the full lifecycle to surface and cu...

#8 📝 Mario Zechner

AI can't cross this line and we don't know why. - Empirically, error in large language models follows power‑law neural scaling relations with compute, model size, and dataset size that form a "compute‑optimal" frontier no model has crossed; OpenAI's 2020 fits predicted those trends and GPT‑3 (175 billion parameters, trained with ~3,640 petaFLOP‑days on a ~10,000‑V100 supercomputer, V100 ≈30 TFLOPS) fell on the predicted line, though some other tasks later show scaling flattening before reaching zero error.

#9 𝕏

Guillermo Rauch unveiled a well-typed useAgent() hook for building fully custom chat UIs (starting with Slack integrations) and teased more ambitious features coming soon.