Bruce J. Hardy

@droidbuilds And now with a visible change to the GIT in VS Code, now I have full access to the file so I can make other changes to the file, all because I humbled myself and allowed the AI do the search for me.

@droidbuilds Hmmm.... complex app, haven't looked at the code in a few months. I've frankly forgotten where and in what file the div actually exists. Why on earth would I waste time trying to figure this out when the AI can search for it and fix it more quickly??!!

MrBeast plans to trap 1000 vibe coders in a room without Claude first person to center a div manually wins $1 million

@Tubes @Ric_RTP Or fire them. As a competent engineer, you should be able to learn these things yourself. Too many engineers are effectively just "vibe coding" right now, and this IS extremely expensive,

@brucehardy1971 @Ric_RTP Guess all these these companies will have to spend billions to train them. Or, you know, just let them do their jobs with the training and expertise they already had without the need for AI.

Microsoft just banned its own engineers from using AI. The tool was literally costing MORE than the humans it was supposed to replace. They lied to you about AI adoption and now the whole narrative is blowing up: Microsoft gave thousands of engineers access to Claude Code six months ago and encouraged them to use it. Engineers loved it and adoption exploded. But then the invoices arrived. Token-based pricing means every query, every code review, every debugging session costs money. At scale across 100,000 engineers, the numbers became so large that Microsoft issued an internal order to cancel nearly all Claude Code licenses by end of June and force everyone onto their own cheaper tool instead. The company that invested $5 billion in Anthropic just told its own people to stop using Anthropic's product because it costs too much. Uber's story is even worse... Their CTO Praveen Neppalli Naga told The Information that the budget he planned for the full year was "blown away already" by April. Uber had rolled out Claude Code in December 2025. By March, 84% of their 5,000 engineers were using it with 70% of all committed code coming from AI systems. Heavy users were burning $500 to $2,000 per month each. Naga himself spent $1,200 in a single two-hour demo session. The company had even built internal leaderboards ranking engineers by how much AI they used. They literally gamified the spending and then ran out of money. Now look at what Nvidia's own VP of applied deep learning Bryan Catanzaro said to Axios last month. Direct quote: "For my team, the cost of compute is far beyond the costs of the employees." This is a VP at the company that SELLS the chips saying that using AI is more expensive than paying humans. Think about what this means for the entire AI narrative. Every CEO on every earnings call for the past two years has said the same thing: AI will make us more efficient, reduce headcount, and cut costs. The stock market rewarded every company that said it. Fired workers, stock goes up. Announced AI adoption, stock goes up. But the actual companies deploying AI at scale are discovering the math doesn't work. The MORE employees use AI, the HIGHER the bill. Goldman Sachs forecasts a 24x increase in token consumption by 2030 as companies adopt AI agents. Gartner just published a report showing that even though individual token prices will drop 90% by 2030, total enterprise AI costs will go UP because agents consume exponentially more tokens per task than basic tools. Meta built an internal dashboard called "Claudeonomics" to track which employees use the most AI. Amazon started pushing engineers to "tokenmaxx," their internal term for consuming as many AI tokens as possible. Both companies are spending hundreds of billions on AI infrastructure this year alone. And Microsoft, the company that bet its entire future on AI, just told 100,000 engineers to stop using the tool they liked best because the per-token bills got out of control. The companies building AI are telling investors it saves money. The companies using AI are finding out it costs more than the humans it was supposed to replace. And even the company that makes the chips just admitted it through its own VP. This is the gap nobody on Wall Street is pricing in. $725 billion in AI infrastructure spending this year across Big Tech. And the first companies to actually deploy these tools at scale are already pulling back because the economics don't work. What do you think?

@bbeliveau @Ric_RTP In any event, what you are describing is not what Microsoft is complaining about.

@brucehardy1971 @Ric_RTP and it costs openai like 10x your subscription price to deliver those results to you - they're still in the first phase of the drug dealer grn strategy

@bbeliveau @Ric_RTP That depends on you!! Have you learned how to control context and state so that you are minimizing token usage? I am at 50% now precisely because of the system I am currently building. It will drop to about 10% after that. Overall the cost will drop as AI matures.

@Ric_RTP My current workflow uses 25% of the daily maximum during codex's 2x promotion in May. That is 50% after elimination of promotion in June.

@Ric_RTP People have not learned how to use AI correctly. I will happily work for a company and pay for my own ChatGPT/codex at $136CAD per month. This is more than enough to complete any task if you know what you're doing while making yourself significantly more efficient!

@haider1 No, the ai creates better code on average than your typical programmer, but you need to have the skills yourself to constrain it. Left alone, it takes the path of least resistance.

Creator of C++, Bjarne Stroustrup: AI-generated code isn't ready — it generates more bugs, more bloat, more security holes, and is nearly impossible to validate "senior developers are already retiring rather than deal with it" The problem is that even a small prompt change can shift the entire codebase in unpredictable ways

@trikcode You refactored 28 files all in one shot??? Using AI does not eliminate the necessity to use your brain first and think carefully about you're doing. Did you use the AI to carefully audit the code base and then generate a safe plan of attack, and then verify the plan and audit?

i asked claude opus 4.7 to refactor a large codebase. 68 minutes, millions of tokens burned - it finished nothing worked. app completely broken

@Adversegroup @TVanWelsenes A review of a json conversion of one 400 line markdown file, converted by Codex, used 22% of my session limit on Claude. The prompt for Claude to complete this review was created by ChatGPT. I did the same test in reverse with codex, and used 2%.

@TVanWelsenes If you have rate limit issues with Claude, just swap to Codex for heavy work and have Claude review it. Works fine for me and I rarely hit any limit

Convince me why I should switch from Claude Code to Codex👀 Am curious about making the switch.

@TTrimoreau Cancelled Claude today when a single prompt generated by ChatGPT ate 22% of my session limit. The same prompt on codex used only 2%, and it did a better job. Having said that, I bet Claude IS better for vibe coders.

Anyone here cancelled their Claude Code subscription and moved to Codex?

@MalikUDraz @KaiXCreator Indeed. That's exactly what codex DID do. It did a better job with 10 times the session limit efficiency ( actually 5 times - we must account for 2x bonus codex is offering for the remainder of May)

@brucehardy1971 @KaiXCreator The cost metric I care about is useful completed work per hour of babysitting. Token/session limits matter, but only after the agent can reliably close the loop.

If Claude Code can now write code, run the app, debug errors, review pull requests, and even fix bugs automatically what exactly is left for junior developers to do?

@adxtyahq Adapt! This AI is interesting stuff. Use the brain power that got you into your jobs to adapt to this new world, so you can apply AI more safely and effectively than those who lack long experience and education.

GitLab layoffs today. feels like we’re entering the era where execution matters more than job titles the people who can build fast, automate, and ship real systems to production will survive this shift. adapt fast or get left behind.

@AYi_AInotes Markdown is still ideal for canonical AI-readable authority/state. HTML is the projection/UI layer humans actually operate through. The future is structured governance → generated operational interfaces, not replacing governance with rendered views.

Claude团队的工程师，已经彻底抛弃Markdown了。 不是Markdown不好用， 是AI变得太快，它已经跟不上了。 以前AI写10行笔记，Markdown刚刚好， 现在AI能一次性输出1000行计划、复杂流程图、完整代码审查， 密密麻麻的纯文字墙谁有耐心看得完？ 作者自己都说，他从来没完整读完过100行以上的AI生成MD文件。 更要命的是：现在都是AI写，我们只看不改。 Markdown最大的优点“易手动编辑”，现在已经彻底没用了。 而HTML，才是AI时代真正的沟通语言， 它能做到的事，Markdown想都不敢想： • 直接生成带颜色的表格、SVG流程图、可点击的原型 • 加滑块调参数、拖拽排序任务、实时预览Prompt效果 • 改完一键导出成代码或Prompt，喂回给AI继续迭代 • 发个链接别人点开就能看，不用下载任何工具 作者直接放出了20个现成示例： 从代码审查的彩色diff， 到可拖拽的任务看板， 从动画参数调试器， 到一键生成的幻灯片。 每一个都是能直接用的生产力工具。 最爽的三个用法，现在就能抄： 1. 代码审查：让AI把PR生成带注释的彩色diff+模块调用图 2. 做计划：生成带时间线、风险表、流程图的交互式项目页 3. 临时工具：让AI写一个Prompt调参器，改完直接复制结果 当然它也有缺点： 多花一点token，生成时间长2-4倍，版本控制不如MD干净。 但作者说：体验提升了10倍，这些代价完全值得。 本质上不是格式之争，而是人机协作方式的升级。 因为Markdown是给人写给人看的， 而HTML是给AI写给人用的。 随着当AI越来越聪明，我们需要的不再是文字墙，而是能互动、能操作、能思考的界面。 现在打开Claude，输入“帮我做一个HTML文件……”，你会打开一个全新的世界。

@elonmusk I disagree. ChatGPT corrected what it believed was your "bad" English and provided the most probable answer. You prompt is meaningless otherwise. It concluded what you really wanted was to count an additional 10, starting at 11. Grok failed to read between the lines.

Not bad

@Layton_Gott I actually use ChatGPT for this discussion we then create a state machine. The next ChatGPT conversation ingests the state machine documents. It follows my plan step by step, prompting Claude and enforcing the results, verifying all tests have been run properly.

@brucehardy1971 Yep exactly 🤝

STOP prompting Claude Code… Instead use this insane workflow: 1. Open Claude chat. Explain the entire project 2. Let it ask questions and push back on assumptions 3. Have it write the first prompt for Claude Code 4. Paste it in. Copy the output back into the chat. 5. Discuss what happened. Any bugs, any changes. 6. Let Claude write the next prompt. 7. Repeat. You end up with prompts that are still yours, but fully polished and detailed by the same system that's going to run them. Tokens are cheap… wasted hours aren’t.

@heygurisingh I don't write my own prompts. I have a conversation and discussion with ChatGPT who is acting as the enforcer. It then generates an extremely detailed natural language prompt for Claude.

Prompt engineering is dead. Nobody writes to chatbots in english anymore. The people pulling godlike outputs are writing prompts like code. It's called json prompting. Once you see it, you can't unsee it:

@housecor Who is enforcing the tests? I have one AI force the other AI to audit the system and then generate the relevant tests. The other AI checks the work of the other through several iterations and improvements. One AI rejects the work of the other AI unless tests pass.

If you think having too many tests is a burden, try coding with too few tests. I've seen both extremes and I'll take the former all day. Especially in 2026. AI writes the tests. AI updates the tests. The tests are guardrails so AI can validate its work.