This sounds like the sales pitch for the AI Prime Ministers in Ray Naylers excellent new book, Where The Axe is Buried.

https://us.macmillan.com/books/9780374615369/wheretheaxeisbu...

I’ve got a feeling that if you looked at the average persons average internet usage it would not be full of interesting stuff.

The world is full of people who find things interesting that I don’t but that’s a fact about me, not the world.

I’ve been paying for Kagi for about 18 months and it’s pretty much the only subscription I have that it’s never crossed my mind to cancel it.

My strategy is generally to have a back and forward on the requirements with the LLM for 3/4 prompts, then get it write a summary, and then a plan. Then get it to convert the plan to a low level todo list and write it to TODO.md.

Then I get it to go through each section of the todo list and check each item off as it completes it. Generally results in completed tasks that stay on track but also means that I can stop half way through and go back to the tasks without having to prompt from the start again.

Cursor has gone to the next level with Gemini 2.5. The reasons it gives for what it’s doing are well thought through and far more in context.

Gemini seems to now advise you when you’re telling it to do something that may not make sense - first time I’ve really seen a non-Yes Man LLM. It’s more like a Yes-but-are-you-sure man.

Cursor in agent mode + Sonnet 3.7 love nothing better than rewriting half your codebase to fix one small bug in a component.

I've stopped using agent unless its for a POC where I just want to test an assumption. Applying each step takes a bit more time but means less rogue behaviour and better long term results IME.

Sounds like a human colleague of mine

> love nothing better than rewriting half your codebase to fix one small bug in a component

Relatable though.

Reminds me of my old co-worker who rewrote our code to be 10x faster but 100x more unreadable. AI agent code is often the worst of both of those worlds. I'm going to give [0] this guy's strategy a shot.

[0] https://www.youtube.com/watch?v=jEhvwYkI-og

If you stopped using agent mode, why use Cursor at all and not a simple plugin for VSCode? Or is there something else that Cursor can do, but a VSCode plugin can't?

Can this be used on AWS RDS? I’ve seen a few things like this that would be great to use but without RDS support they’re unusable for us.

No, not until AWS decides to add it as a supported extension, which will most likely never happen (you can see the current list of supported extensions here: https://docs.aws.amazon.com/AmazonRDS/latest/PostgreSQLRelea...)

Building a RN app without any features (authentication, notifications) is easy - but adding those features and then navigating the outside-app ecosystem isn’t. It’s definitely not like the web.

Strange that LlamaParse is mentioned in the pricing table but not the results. We’ve used them to process a lot of pages and it’s been excellent each time.

> The core issue, however, is the inability to participate in the actual craft. Design decisions are buried in React components with cryptic expressions like flex items-center shadow-lg p-6 hover:bg-gray-50 dark:bg-gray-800 py-[calc(theme(spacing[2.5])-1px)]

Give me Tailwind standardised utility classes over having to search through the codebase to find what a dev 6 years ago called a CSS class.