Ralph

Hello, can my friends in China see the messages I’m posting?

OpenAI is working on a new "Translation Block" widget in ChatGPT Fun fact - one of the supported languages is "High Valyrian", a fictional language created for Game of Thrones

This is why half of the “AI Agent” demos online feel fake. They show one tool call and act like the whole system is solved. But the real pain is orchestration, context, retries, permissions, and making the final answer not look like garbage. Chart is not perfect, especially around MCP, but the direction is right.

A lot of people still describe AI Agents as just “LLM + tool calling.” But once you actually run the workflow, you realize it’s not that simple. The user query is only the entry point. Behind it, there’s memory, context retrieval, task decomposition, skill routing, MCP/tool calls, execution, result collection, and final response generation. I broke the full chain down into 20 steps.

@esimuse ChatGPT全能，克劳德比较注重代码使用性

现在gpt、gemini和claude那个最好用呢？

为什么你越来越容易愤怒？

The crazy part isn’t that GPT-5.2 is outdated. The crazy part is that “outdated” now means six months old. AI timelines are basically dog years now.

You can’t even pick one good stock from thousands of U.S.-listed companies, but you think you can find “the one” among 8 billion people.

The Real Reason You Struggle With Learning

@esimuse 能用肯定用英国卡啊，香港IP现在不太行的

🇬🇧 英国卡 vs 香港卡，到底怎么选？🤔 别一上来就问哪个更高级。 卡不是用来装的，核心就看你拿它干什么。

@esimuse 确实是这样的，有一张国外的卡会比较安全一点

+86的号吗包涵了太多的信息了，而且容易风控，何必一直纠结于用这个呢？用一张国外的卡保护隐私不是更好吗？

The scary part isn’t that it improves productivity. It perfectly captures the four things I say to AI every day: Fine. Not fine. Always fine. Listen to me.

让ai自己做项目何尝不是一种ntr呢🤣🤣🤣

Chinese developers might be the most cybersecurity-aware people on earth. I’ve never seen one visit GitHub from their real IP.🤣🤣

@gregpr07 Model self-identification is probably the least reliable eval we have, but it’s still the funniest one.🤣🤣

I think I know why deepseek is so good

Meta’s AI shift is more interesting than it looks. People keep asking: “Will AI replace workers?” But maybe that’s the wrong question. Meta reportedly moved thousands of employees into AI-focused teams after major layoffs. So the real question may be: What happens when companies don’t just use AI… but start reorganizing the entire company around it? AI is not only changing products. It is changing job titles, teams, budgets, and who still matters inside the company. Maybe the future of work is not “AI replacing humans.” Maybe it is humans being reassigned into whatever AI needs next. That part feels much less comfortable to talk about.

AI labs are entering a weird phase. For years, everyone was asking: “Who has the best model?” But that question feels incomplete now. Because if every frontier lab needs insane compute, insane talent, and insane capital just to stay in the game… then the real question becomes: Who can turn intelligence into profit before the cost of intelligence kills them? Maybe the next AI war won’t be about benchmarks. Maybe it will be about margins. And honestly, I’m not sure most people are ready to talk about that part yet.

@dhh This is the part people underestimate. AI doesn’t just save typing time anymore — it changes what scale of refactor even feels possible. The hard part is keeping review quality from collapsing.

For complicated agent work, it's amazing how much GPT5.5 has improved. I found 5.2 to be very far behind Opus. Now using Opus 4.7 after 5.5 feels like a big step backwards. Gotta love this level of competion! Strong comeback for OpenAI.

@CrossCheckDev yeah this is the part people underestimate. the pain is not one big task. it’s 200 tiny switches between tools all day. if agents can quietly remove that, people won’t care what we call them.

@RalphJing_ The real ROI isn't chatbots, it's eliminating the context switching tax between systems. Most teams are bleeding thousands of hours on connective tissue work that agents could handle silently while everyone chases the flashy demos.

I’m not sure the “agent boom” is going to look very cool. Maybe it’s not some perfect AI coworker with a name and a face. Maybe it’s just boring stuff getting automated quietly — tickets, docs, CRM notes, internal reports, code review, all that messy work nobody wants to touch. Which is kind of funny. People keep selling agents like they’re new teammates, but maybe what workers actually want is much simpler: less annoying software. not sure. feels like that might be the real market.