Today I'm sharing some ideas on how to turn methodologies into skills. Often, after reading a book or watching a video, I feel like I've learned something, but I can't fully apply it to my own projects, wasting a lot of inspiration.
At the investor conference, Liang Wenfeng systematically explained DeepSeek's underlying logic: the company takes AGI as its sole objective, driving management through vision rather than KPIs; it adheres to open source and a restrained pricing strategy of "recovering costs in ten months," believing that the more restrained, the easier it is to succeed...
The GPT 5.6 series was recently released, and it's been fiercely competing with Fable 5 on various benchmarks. Everyone wants to know which one is stronger. I also want to know. Instead of just looking at the benchmarks, let's use them directly, put some cases together to test these two models, and see which one...
Writing short articles and polishing copy can be done by many AI tools these days. However, for more complex professional scenarios, such as managing an enterprise's internal knowledge base or creating long-term, in-depth content, adding new material weekly...
At the recent World Artificial Intelligence Conference, the ScienceOne team from Zhongke Wenge unveiled two trump cards: S1-Omni, a unified scientific multimodal reasoning model specializing in deep understanding, prediction, and generation in scientific scenarios; ...
Last night, the Dark Side of the Moon quietly launched its new flagship model, the Kimi K3, and my WeChat Moments and social media groups were practically flooded with it. This time, the Kimi K3 not only doubled its specifications, boasting a 2.8T processor, but also directly broke global open-source benchmarks...
Today I'm sharing some things suitable for GPT 5.6 Sol reverse proxy. Reverse proxy can be understood as converting the client's model into an API for use. I used to think the official team didn't want users to do reverse proxy, but...
Today I'm sharing some things suitable for personalizing Codex. Codex is very user-friendly, but the interface is basically determined by the official documentation. I open it many times a day, and after a while, it feels like something's missing—something personal. Luckily, I found C... on GitHub...
I uploaded a flat lay image, and five minutes later, the main image for Taobao, the white background image for Amazon, the product recommendation image for Xiaohongshu, the long image for the product details page, and the model video all appeared. It only cost me 9.9 yuan.
Over the past year, video agents have become one of the hottest areas in the AI video industry. Many agents can generate video clips from a single sentence, but when it comes to creating complete video works, the generation process remains crucial...
Tencent's Hy3 recently came out, and WorkBuddy users can use it for free. Everyone's scrambling to get their hands on it, and there's a long queue. Hy3 is an upgraded version of the Hy3 preview released at the end of April.
Actually, if you want the capabilities of GPT Work + Codex, there are already mature and affordable alternatives available in China. In my opinion, the best experience currently available, arguably the best domestic alternative, is Baidu's Baidu Partner. It doesn't require environment setup, and it's not affected by region...
The upcoming release of Seedance 2.5 has attracted industry attention, but it also comes with an extremely high price tag. Rumors abound that a single AI-generated short drama can cost tens of thousands of yuan, making creators incredibly wealthy. However, this only applies to those who actually write prompts and run the scripts...
Today I'm sharing some things suitable for voice replication. To be honest, the most annoying thing about doing AI voice work right now is that you only want to change a few words in the generated audio, but you have to re-record, re-dialogue, and re-edit it.
Today I'll share some tips for making AI agents output more concise. Normally, when we use AI agents, the output, regardless of its accuracy, often includes a lot of rambling and unnecessary information, which is a waste of our time and energy to read. Using them...
Today I'm sharing something suitable for combining two powerful tools. There's always been a debate about whether Claude Code or Codex is better, but we're all adults, why make a choice? I want both.
Today I'm sharing some tools suitable for local AI video creation. To be honest, current AI video generation technology is quite mature, with only one drawback: it's expensive. Pricing is basically per second, and it's difficult to guarantee a perfect take; my wallet really can't handle it...
In today's fiercely competitive e-commerce market, a high-conversion main image often determines a product's survival. Traditional visual design processes are lengthy, costly, and heavily reliant on specialized skills; a high-quality poster can easily cost thousands of yuan...
Today I'm sharing some comparative tests of open-source and closed-source models. I've selected four open-source models: Kimi K2.7 Code, GLM 5.2, MiniMax M3, and DeepSeek V4 Pro, to compare with Sonnet 5 and see how Sonnet 5 performs...
Claude has recently started suspending accounts on a large scale again, both new and old, without any warning. The most frustrating thing is that all the context, personal preferences, and project workflows that were painstakingly built up beforehand have to be completely reorganized and migrated...
Today I'm sharing some things suitable for having a desktop pet. When we use Claude Code or Codex CLI, the text in the terminal is constantly scrolling, and unless you're paying close attention, you have no idea what the agent is doing. Today I'm recommending something for...
Today I'm sharing some tools suitable for video/image face swapping. When I want to create funny videos or images, face swapping with Photoshop always ends up looking bad, and AI is a bit expensive. I finally found something great on GitHub called...
You might also be frustrated by this: AI-generated blockbusters have incredibly high-quality visuals, but the moment the characters start speaking, it instantly pulls you out of the story. The scene is clearly about a life-or-death situation, but the voiceover sounds like someone calmly reciting a product description...
Today I'm sharing some things suitable for web design. When we use agents like Cursor, Claude Code, and Codex to create pages, we often encounter a problem: we tell the AI what style of page we want to replicate, but...