Owen Lee

I really tired of fighting jq C bindings in Python. so I built purejq — a pure Python jq implementation that's 6–40x faster than the C bindings, no JSON serialization round-trips. Runs anywhere Python runs: Pyodide, Lambda, sandboxes, you name it. 96% of jq's test suite passes. github.com/adam2go/purejq

@sunwukong_DEX 来啦，关注一波～

师父说让俺关注几个蓝V 👀 让俺看看谁还没被俺关注，评论区留言 👇

@kaialan__ @X demo

@kaialan__ @X Hi Serena, I build Tilo, AI-native SaaS agents with the ROAM Loop: Render, Observe, Act, Memorize. github.com/adam2go/tilo-f…

looking to connect people on @X if you're into - design - building SaaS - vibe coding - AI tools - shipping in public - figuring it out as you go say Hi or drop what you're working on looking to follow active ones 👋

@GithubProjects Let's Connect👋👋 github.com/adam2go

Do you have a GitHub profile? Share the link and let's connect.

GLM 5.2 vs DeepSeek V4: share of tokens on OpenRouter GLM 5.2 has had a very fast uptake relative to most open-weight launches

@opencode GLM-5.2 is actually really good, honestly. Great price, great performance, and I've already integrated it into my product.

we've added unique user rankings some models are token heavy so they skew upwards in rankings - unique people using the model is a more accurate ranking we'll orient more of our data around this metric

@wong2__ 本地模型适合解决小型、简单场景任务，但复杂、重推理的任务还是需要云端更先进的模型。而且模型迭代速度太快了，本地模型用不了多久还得换

感觉我们离这样一个拐点不远了：当本地开源模型足够强，硬件成本低于长期云订阅成本时，重度用户会开始从“租用 token”转向“拥有算力”。就像历史上从大型机到PC的转变。

8 hours a day collaborating with Claude + ChatGPT. The rest of the time? Thinking.🤔 Tools handle the execution. Humans own the direction.

A good rule of thumb for preventing AI slop (in writing, design, etc): Is your input (the context) longer than the output? I've found that for the AI to produce quality results, my input is often 3-5 times the length of the output. If your input is much shorter than the output, it's almost certainly going to produce slop

@Shayueth 来喽

最近中文圈蓝V互关的热度又回来了！ 说实话，圈子不大，大家抱团取暖真的很有必要。 #蓝V互关

I'm increasingly finding my creativity insufficient because AI can realize my ideas so quickly! Previously, product design involved a lot of time for designers, software engineers, and testers. I can handle the entire process myself now, but I don't have much time for thinking🤣

@massiwak125895 互关👋

（X 涨粉 / 互关方向） 分享一个冷知识： 想快速涨粉，别只发帖等别人来。 直接去大 V 的互关帖评论区和关注列表，一个个点过去。 那里才是真正想互关、正在起号的人。 别被动等待，主动出击，涨粉会快很多。

@Levi_1203 诚信互关一波👋

好久没 #蓝v互关 了 放心大胆的来和我互关吧！

104subagent, 18mins, token is burning🤨

@zarazhangrui Strongly agree with "Doing everything means doing nothing. "

Every other product right now is "an AI agent that does everything in your work & life & integrates with everything." Cool, that's just Claude/Codex. If you want me to use your thing instead, it needs an opinion & a soul. Build small & sharp, not big & generic. Doing everything means doing nothing. Speaking to everybody means speaking to nobody.

AI 不是人的替代者，而是人类主体性边界的放大器。它会放大人的能力，也会放大人的逃避，会拓展人的心智，也会暴露人的空洞，会成为工具，也会被视为“他者”。真正的问题不是 AI 会不会像人，而是人会如何在 AI 的参与下重新成为自己。

Render → Observe → Act → Memorize This is the direction I believe AI-native SaaS is heading: not chatbots wrapped in dashboards, but agents that can render, observe, act, and remember. GitHub: github.com/adam2go/tilo-f…

In recent versions, Tilo added: v0.5: durable conversation runtime v0.6: hardened runtime primitives + multi-app examples v0.7: ORID-based context reflection GitHub: github.com/adam2go/tilo-f…

Beautiful machines in space