Cursor 2.0 can now: 1. Use any frontier model or our own agent model 2. Run multi-agents with built-in git worktrees 3. Create plans with one model, implement with another 4. Review + create a PR in one click 5. Close your laptop and handoff to the cloud What should we do next?

How to monetize the same content 5x as a Digital Writer: • Write a paid newsletter ($) • Publish snippets behind Medium's paywall ($$) • Group newsletters by topic & publish as a book on Amazon ($$$) • Turn all this content into a course ($$$$) • Advise people 1:1 ($$$$$)

Dudes who keep tweeting with 0 likes fear nothing. Respect.

How to actually get things done: no meetings by default  quick huddles on Slack  async work in Notion docs

What % pay cut would you take if you didn't have to attend any calls?

$0 MRR. 100% transparency

Make something before you say something

We’re in a slop bowl recession

We taught AI Meeting Notes to cite its sources.¹ ¹Meeting summaries now include citations → jump to where each takeaway came from in the transcript.

Why the hell did @danshipper launch a new dictation app? There's already SuperWhisper, Wispr Flow, VoiceInk, the built-in Mac tool, etc. (Well, the built-in Mac dictation tool is garbage, but you get my point.) I asked Dan about it. He basically said: If we like it, chances are our readers will too. That’s how we build everything. Dan’s an oddball, but maybe he's the future. Monologue, his dictation app, has a few quirks. Scifi design. Multilingual. Auto-editing. But he didn't make any of those decisions based on some kind of market demand or need. When I pouned him with questions about how he decides what to build, he said: When one person can build something great in days, you don’t have to build for millions anymore. You can build for people who share your taste. It reminded me of what happened with podcasting. Before podcasts, radio had the same generic news shows. Then podcasts started getting more niche-y and focused. - You could do a startup podcast - Then just for startup design geeks - Each niche is built on the host’s obsession Mixergy was based on my personal obsession with "TELL ME HOW YOU DID IT!" That personality-based creativity is happening in software right now. AI lowered creation cost/time. Taste became the draw. It seems to be working ok for Dan. People like @nateliason can't shut up about Monologue on X. I published my conversation with Dan on YouTube. 17 people already listened to it! You can be the 18th, here: https:// youtube.com/@TheNextNewThi ngAI?sub_confirmation=1 …

I'm stuck. I sold 13 ad spots, but TrustMRR displays 12. It's because the buy button is hardcoded. I manually push sponsors to prod for now. I didn't know the app would receive that many visitors in such a short time (35k visitors in 16 hours). How would you handle this? Possible solutions: 1. Add two ad spots. 2. Refund the last customer. 3. Rotate ads like Times Square (two ads per card, flip them). 4. ??? I would love to know what you think before making the next move. Please hit reply!

Moved from AWS → Render → Hetzner + Dokploy. AWS: $100/mo, 3 services, IAM & networking hell, surprise bills etc. Render: $40/mo for very low/basic specs, meh control (but still nice to work with) Hetzner + Dokploy: $5-10/mo, full VPS, full control One-click deploys. Traefik reverse proxy. Auto SSL. Postgres/Redis included. GitHub Actions ready. Same app. 95% cost reduction. 10min setup. Actually fun. Next step: Raspberry Pi in my closet. Then I’ll grow up and over-engineer everything with K8s like a real adult

Bye Bye fake DR screenshots I built http:// awesome.tools/leaderboard, the database of verified SEO traffic sites! 1. Submit your url: https:// awesome.tools/submit 2. Get a full human-written review 3. And a nice dofollow backlink as a bonus There's no way to game the system. It's all running on my server, and users can't edit the SEO data. + It comes with the leaderboard of startups with the most clicks & SEO traffic in the last 30 days Go get verified!

The world runs on TypeScript & JavaScript. Our bet is that AI engineering will follow suit. The growth in @aisdk downloads and adoption has been astonishing. When we wrote the Ship AI keynote it was at 3.4M weekly downloads. A couple weeks later, it’s now at 4.1M https:// vercel.fyi/ai-npm

The best customers are cash rich, time poor, and in big pain.

Vercel Agent has a new skill: incident response. It can now intelligently investigate anomalies to surface critical issues and speed up time to resolution. New Agent users can claim a $100 free credit.

Cursor now has a built-in web browser. You can say "start my app" and it will run the dev server in a terminal and then navigate the browser to localhost:3000. It can take screenshots of the page for iterating on design, and you can send DOM elements to the agent with prompts!

We should not try to make AI agents into something more complex than they are. Technically speaking, an agent is a workflow. Many agents are crons, of which workflows are a superset. Because agents are workflows, we’ve seen a renaissance of workflow builders: Zapier, n8n, etc Whether you build an agent visually or with code, underpinning the workflow there must be an engine and infrastructure that makes it reliable. A workflow is a different kind of software than a web server. It’s not request-response. It consists of many steps, can fail a lot and is expected to recover, it can run for a very long time. This is why we’re so excited about @WorkflowDevKit . An agent is just “𝚞𝚜𝚎 𝚠𝚘𝚛𝚔𝚏𝚕𝚘𝚠”. Pair it with @aisdk (where each tool call is an “𝚞𝚜𝚎 𝚜𝚝𝚎𝚙”), AI Gateway for token reliability, and Fluid for efficient compute… and you have all you need to build the future of software.

I love vibe planning

Tuesday: @levelsio tweets Thursday: @marc_louvion ships Sunday: $15K MRR