PageAI turned 1 month old today 🥳. 🤑 $1270 revenue in the 1st month, not bad!. been releasing like crazy since launch. super happy with our stack: . - @nextjs & @aisdk frontend.- @shadcn UI & dashboard.- @Hetzner_Online for long running jobs .- @supabase for Auth & DB.-

4️⃣ Should you try it?. If you are not a Cursor user. 👉 You should absolutely get Claude Code. It'll blow your mind. And I now realize a lot of the hype comes from ppl that first experienced AI coding that works. If you are new to Agentic coding?.This'll be incredible to try.

3️⃣ Cost, value for money & error rate. I got the $100/mo Max sub & got my money's worth in the first 5d. It's good value for money atm + it allows you to use the full context. Should you use the full context?. With current gen LLMs, prob not. But it's transparent about it.

2️⃣ Workflows: new and old. I quickly found myself missing a bunch of QoL I took for granted:.- autofix.- checkpoints.- deterministic type & lint checks.- auto formatting.etc. The freaking bell sound was missing when it was done 😃. Some things are really easy to do and work.

1️⃣ Setup & code quality. I love using the terminal. But writing big prompts, dealing with long path names, URLs & *especially* code blocks just sucks. Onboarding was rough. The terminal window felt buggy & would jump frequently. I couldn't @ a lot of the files, I think it can't.

I talk about my full experience here if you prefer video. 👉 👈.

Claude Code is supposed to make us 10x devs.Should I stop paying for Cursor?. I needed to know. So I used Claude Code for 7 days straight. Only. Claude. Code. Here are my thoughs:

Also: if you wait it out a bit it'll probably get a good UI too. It feels like early days but we can already see a lot of tooling pop up in the open source space . I think that's ultimately what will make Claude Code a huge success.

BUT if you're NOT a Cursor user. 👉 You should absolutely get Claude Code. It'll blow your mind. And I now realize a lot of the hype comes from ppl that first experienced AI coding that works. If you are new to Agentic coding?.This'll be incredible to try.

🔟 Should you use it. I think this architecture is the correct, non-hacky way to build AI tooling. I expect many UIs and tools on top via it's SDKs, so I recommend you get comfy with it. Will it replace Cursor?.This q doesn't make sense. You'll need an IDE eventually, and.

9️⃣ Cost. It's good value for money atm + it allows you to use the full context. Should you use the full context?. With current gen LLMs, prob not. But it's transparent about it. I got the $100/mo Max sub & got my money's worth in the first 5d. People use >$1k worth of tokens.

8️⃣ Error rate. It still produces a lot of errors, like all current gen LLMs. It's generally okay, but not to the point where you can turn on auto-accept and just vibe code. It does a lot of silly mistakes and have mangled files for me many times using wrong encoding or writing.

7️⃣ Running CC as a script & CI/CD integration point. claude -p. This is SO underrated. This is a non-interactive mode that enables Claude Code to be used as a CLI utility. So all of a sudden all your pipelines get AI capabilities. Hugely powerful.

6️⃣ Architecture. You can pretty much emulate all the good things from other IDEs, either through MCPs, rules and/or hooks & scripts. Hooks are deterministic tasks you can run at certain points and they are cracked. The architecture is solid and it comes with a huge bonus:.

5️⃣ Setting up. (yes, you can set up the bell sound w/ hooks). Out of the box Claude Code doesn't assume much about your set up, which makes it great as a generic tool. But that is also its achilles heel. Many things you get out of the box with Cursor you now *need* to set up.

4️⃣ Powerful workflows. Some things are really easy to do and work well.- sub agents.- thinking.- planning.- scripting / using the CLI.- leveraging temporary TODO lists. But I had to do a lot more setup:.

3️⃣ Workflows & quality of life. I quickly found myself missing a bunch of QoL I took for granted:.- autofix.- checkpoints.- deterministic type & lint checks.- auto formatting.etc. The freaking bell sound was missing when it was done :|.

2️⃣ Writing first code. The quality seemed good. It was fast. But I couldn't quite follow what it was doing as it spit out tons of logs. I act stopped watching it eventually :D. Tool calls and diffs weren't observable like they are in Cursor. Felt like it needs more.

1️⃣ First impressions. I love using the terminal. But writing big prompts, dealing with long path names, URLs & *especially* code blocks just sucks. Onboarding was rough. The terminal window felt buggy & would jump frequently. I couldn't @ a lot of the files, I think it can't.

I've spent > 1 week strictly using Claude Code and experiencing it from the lens of a pro Cursor user. 👉 It was quite the rollercoaster.

Is Claude Code really that good? 🤔. finally, an objective take without the clickbait hype: