ojr

you need the talent to come up with a better architecture than LLMs, not sure if it will be coming out of Anthropic, but they have a chance.

It's like the boy crying wolf with these models, Deepseek is not as good as Claude, I like using Gemini Flash Lite even, it is good enough for most of my crud app tasks, brownfield projects you don't need the strongest models, but greenfield projects with lack of context from indexed code and lack of domain expertise, I think the Opus models have been working for people

I agree the tech doesn't matter but React Native helps with the deeply layered problem of execution.

I have a way different taste, my coding agent doesn't run in a shell and proposes file changes first without changing them, so its prompt -> propose changes -> accept, bullet has the typical agent workflow, prompt -> create changes -> revert (if you don't like what was changed)

My coding agent leverages vector search and grep, just using grep takes longer and seems more expensive per task.

harnesses are becoming the new tech stack, I created my own and will compare it to how bullet operates, I’ve also built and deployed iOS and Android apps with my tool. When you have built something with the tool, other than the tool, it’s stronger proof.

IDEs can autocomplete tailwind classes but eventually you just memorize them, also with AI assisted coding I don't even write the classes anymore just small adjustments like changing pt-3 to pt-4, AI implemented dark mode in a small app I am working on in one shot, I think tailwind being in the training data helped tremendously with the task.

the resale market for disk has been on a downtrend for years, you can sign into someone's else psn account too and share games, you are a washed up gamer, its okay I am washed up too.

it just a higher level of abstraction but not high enough that it replaces developers, I don't think the LLMs are even capable of reaching that level. Saying the creative process is gone is laughable. A lot of people make these claims and only have an internal tool that no one uses to show for it.

I built a "simple" storage application recently, uploading large files from a phone is not trivial, you can't just command Claude to do it without giving it good context.

This truly shows how king making in venture capital is done, kids have the MIT pedigree but sometimes this is not enough for certain demographics, give them a ton of money to explore ideas and pivot, product is a vscode fork that sells subsidized AI, only possible with venture capital. Providing inference at unsustainable rate deemed as "product market fit". Product loses money until they exit.

VCs that say "I always knew the team was special", give me a break.

I can use Gemini 3 Flash with the harness I built for around 8 years and still not exceed the cost of a Mac Studio with 128GB, the price for privacy is very high. Agentic flows that get stuck can be worked around but I prefer developer velocity.

I use Gemini 3 Flash, I've seen the Claude Code setups, bullish on Anthropic people are driving up tokens but I am able to produce outcomes with a fraction of the money.

well 2026 is easily the most software I shipped so AI is definitely making the process faster. Sad if you with agree with the post, I'm outlier though AI has enabled me to build several subscription apps that was able to make money, not a lot but it will compound trust me.

I was able to release two new iOS apps including a game, and a cross desktop application just this year. I refuse to go back to writing code by hand. If it doesn't help your productivity that's okay, ignoring how productive it has made developers like myself is a choice.

AI was also able to help me create my first subscription payment workflow.

It is like farming without Roundup, less crops, more energy, less toxic chemical risks.