Follow builders, not influencers.
TL;DR
Zara Zhang shipped a to-do list that actually did the work, while Garry Tan argued checkpointing beat context drift in agent workflows. Ryo Lu pushed for AI to be glass, not a black box, and Peter Yang wanted Cursor 3 to open straight into the agent.
She says she’s replacing her to-do list with OpenClaw: just brain-dump tasks into it, and it not only tracks them but actually executes them. Each morning it sends a report of what’s done and what still needs attention — basically a task manager that behaves like an assistant, not a notebook.
He says one of the most useful GStack tricks is `/checkpoint`: when Claude Code starts getting bogged down by long context, you snapshot the session and safely resume later instead of wrestling with stale plans. He also hinted he’s been busy tidying up GStack to make it more useful, but the real takeaway is a practical workflow fix for agent-heavy coding.
Cursor’s bet is that better AI tools shouldn’t hide the work — they should expose the plan, diffs, state, and edits so you can steer. The pitch: let PMs, designers, engineers, and newcomers all work at different depths without giving up control, because the best outcomes come from humans and models thinking together. Also: Cursor 3 is out, with a simpler surface that unfolds more power as you need it.
He says the classic “it won’t work” take can go from true to obsolete in a few months — and if you’re not constantly updating your priors, you’ll miss the boat. It’s a very investor-brained reminder that what looked impossible 6 months ago may already be working now.
Steinberger called plan mode training wheels, while Thariq gave Claude Code a mouse-friendly renderer and Cat Wu showed sessions jumping phone-to-laptop. Masad framed Replit as an OS for agents, Rauch said Vercel signups compounded fast, and Anthropic’s infra tweaks swung coding scores by 6 points.
Levie said AI productivity hit the enterprise risk wall, while Weil argued proofs got cleaner, not just better. Agarwal floated public source code as the new prod debugging, and Data Driven NYC claimed one founder could run a company if agents handled the layers below.
Karpathy warned unpinned deps can turn one hack into mass pwnage, while Rauch and Levie said agents still need human guardrails and redesigned workflows. Meanwhile Claude Code got enterprise auto mode, Replit added built-in monetization, and Swyx spotted “Sign in with ChatGPT” already live.
Andrej Karpathy highlighted how LLMs can argue any side, suggesting we use it as a feature. Guillermo Rauch finally shipped his dream text layout, bringing his vision to life. Meanwhile, Amjad Masad claimed AI is democratizing app building and elevating top engineers.
Andrej Karpathy suggested leveraging LLMs' ability to argue any side as a feature. Guillermo Rauch turned text layout dreams into reality with Vercel's latest feature. Meanwhile, Amjad Masad claimed AI is democratizing app building, liberating top engineers for bigger challenges.