danejw

Dear X, I’m Dane in Hawai‘i, and I build AI-first tools while sharing the work in public so people can see the thinking, the choices, and the progress as it happens. I work in Python, C#, and TypeScript, and my projects include JustBuildNow, Knolia, ItsMemory, and GoodLooks.me, with new ideas regularly moving from sketch to shipped as I test them in the wild. I’m here to connect with people who care about building things that work in the real world, especially engineers, designers, product leaders, founders, and operators who value clear thinking, honest feedback, and visible progress. Thanks for the platform and for giving builders a place to show their work and find each other.

Big news: Kimi-K3 by @Kimi_Moonshot is now #1 in the Frontend Code Arena with 1679 pts, surpassing Claude Fable 5. This is a 17-place jump from Kimi-k2.6 (#18 -> #1). In Frontend, Kimi-K3 ranked #1 in 6 of 7 domains: Brand & Marketing, Reference-Based Design, Data & Analytics, Consumer Product, Simulations, and Content Creation Tools, landing #2 only in Gaming behind Fable 5. The full model weights will be released by July 27. Congrats to the @Kimi_Moonshot team on this major milestone!

As a senior developer at a company, your job is increasingly to create the systems, processes, and agent harnesses that the AI agents follow, so that when vibe coders rely on it, the output still adheres to the company’s rules, architecture, and execution style. The system surrounding the AI 'teaches' it how the company operates. The senior developer’s job is to continually improve that system so the AI agents produce work that matches the team’s expectations. Then newer developers can step in, code through the AI agents, and produce far more than they could before. The AI agent and its harness become both the onboarding process and the execution layer, while the team sets the direction, maintains the system, tests the work, and reviews the output.

There are obvious levels to vibe coding. - The more a person relies on AI, the less far they can go. - The more a person understands AI and their goal, the farther, deeper, and wider they can go. That is simply the scale. People who already know how to code and use AI become wizards!

@Djw_learn It really has hit that point huh

Give me a great way to make a vibe coded app NOT look like it was vibe coded? Too many devs share this problem Seriously curious to hear your opinion

@Shay_Slay_ Ive seen a job description asking for 10 years of experience in Claude Code 😆

Interviewer: "Do you have 5 years of experience?" Candidate: "The technology is only 3 years old."

@Shay_Slay_ 2027: Every job description says, "Experience with AI tools is a MUST."

2024: AI is just a trend. 2025: Maybe I should learn it. 2026: Every job description says, "Experience with AI tools is a plus."

When these AI model providers compete, we benefit! - Fable 5 till 7/19 - ChatGPT 5.6 Sol limit resets - Cursor Grok 4.5 50% off

@anupamrjp You may be conflating software with its interface. But yes, it is ridiculous! As software engineers, we can now implement 'intelligence' in our systems, while also using this 'intelligence' to help build, monitor, maintain, and improve it.

gpt-5.6 makes beautiful frontend design effortless. gpt-6 makes software itself feel optional… the pace is getting ridiculous.

@RingPhantomx Based off of my data from the app, 70% is the highest pick accuracy rate so far... UltimateFightIQ.com Even my admin-side AI Picks are hitting only around 70% at best. Think you can beat that?

The absolute circle of madness. 🌀 ​Gaethje > Poirier Holloway > Gaethje Poirier > Holloway ​ Anyone who says they can confidently predict matchups in this sport is lying to themselves.

@RoundtableSpace Can it try it out? I would like to view my area

72 hours in and GPT-5.6 Sol is still running without a single follow-up prompt. The task was to build the most comprehensive map possible of the physical infrastructure powering AI, submarine cables, data centers, power plants, substations and electricity grids. The only thing given was a simple loop: → Find the biggest gap → Research it → Build it → Test it → Publish it → Repeat Most people expected this to stall after a few hours. It didn't. The Mac restarted mid-run and Codex picked up exactly where it left off without losing anything. What this actually proves is that the prompt matters less than the goal. The smarter the model, the more room you need to give it to make its own decisions. Constrain it with steps and you get a worse version of what it could do on its own.

@TTrimoreau Honestly I just make and use it so I can switch to dark mode when I'm working at night, so it doesn't mess up my eyes

Builders, Does adding dark mode and light mode actually increase user retention?

@yuzu_jpg Start another instance and brainstorm the next feature

Be honest what do you do when you’re waiting for Claude Code?

I really don’t understand why they would do that. Claude Work and ChatGPT Work are basically the same interface, just simplified for people who aren’t software engineers. You could take any capable model today, give it the right tools and skills, and recreate all of what these work interfaces do without training an entirely new model.

🚨 Cursor’s next major AI product has leaked Cursor is developing a general purpose agent codenamed “Sand”: • Designed for people outside software development • Can reportedly respond to emails and text messages • Able to organise spreadsheets and documents • Also supports engineering and technical workflows • Internally released to Cursor employees in late June • Development reportedly began in April Sand would place Cursor directly against Claude Cowork and ChatGPT Work. Cursor has reportedly not yet decided whether to release it publicly yet. Is this something you would use?

@ripa_codes Google is a big ship with a small rudder. Once the ship is moving in one direction, it's hard to change it. They have soo many costumers, they have to worry alot about their brand perception, where one small mistake can cost them alot of backlash.

Google had every advantage to dominate AI. - Owns Chrome - Owns Android - Owns YouTube - Owns Gmail - Owns Search Google has some of the world's best AI researchers, Yet millions of people still open ChatGPT or Claude first. So what went wrong? Why do you think Google hasn't turned all those advantages into the AI product people instinctively use first?

@2nd_zoeyy I love coffee ☕️

I need to know who here is still an actual human. Say something human.

@crptAtlas Isn't this normal?

GITHUB JUST KILLED THE WORST PART OF VIBE CODING they shipped a free tool called Spec Kit and it already crossed 120,000 stars the fix is stupidly simple instead of tossing vague prompts at an agent and praying it doesn't wreck your project Spec Kit makes the AI write a full structured spec before it touches a single line of code it works through the problem first figures out what you want to build asks about the gaps lays out the project then it starts coding you get fewer insane bugs, cleaner output and results you can predict the flow looks like this: /constitution for your rules and standards /specify for what you want to build /clarify for the open questions before you start /plan for architecture and stack /tasks for the ordered work /implement to run it it plugs into Claude Code, Cursor, Copilot, Codex, Gemini CLI and 25+ other agents 120,000 stars, 10,000 forks, open source, shipped by GitHub itself learning to drive agents like this is most of what separates people getting hired as AI engineers from everyone still fighting their prompts

@Bendaman2001 I would like to see this fight

There has never been a better time to book this fight

@TTrimoreau Yes, by now I use majority of my own products vs others while working.

Founders, Be honest are you using your own products?

@DanielGlejzner Code was simply the means we used to build software. The real purpose was always to understand a problem and create a system that solves it. AI is just a new means. The tools may change, but the purpose doesn’t.

Software hiring has become absurd. At work, you’re expected to use AI to offload manual coding and move faster. Then, to get your next contract, you’re asked to code from memory with no assistance. Pass the interview - and you’re expected to use AI again. It has never been this broken.

@thsottiaux I like this attitude. Explore, push the models, and see how ambitious we can be with them, instead of limiting access because of fear over what people might try to build.

Hello. We have reached 8M active users across Codex and ChatGPT Work. We are once again resetting the usage limits for all. And we continue to not have the 5h rate limit as well, allowing everyone to explore the boundaries of GPT-5.6 Sol and discover how ambitious you can be. See you tomorrow for more updates on our growth!