Jeongho Nam

Thank you @AnthropicAI I also got it by github.com/samchon/typia

We built an AI that writes full backend apps with 100% success rate. Then we broke it on purpose. The code was unmaintainable — no shared modules. Change one field? Regenerate 50 implementations. So we rebuilt everything. Success rate crashed to 40%. How do you find bugs you don't know exist? Throw weak models at it. A 30b model exposed every schema ambiguity that GPT-4.1 papered over. qwen3-coder-next raw function calling: 6.75%. With validation feedback: 100%. We shipped builds with NO system prompt — output quality was indistinguishable. Types beat prose. Back to 100% with GLM v5. - Detailed Article: dev.to/samchon/autobe… - Github Repository: github.com/wrtnlabs/autobe

I'm currently working on four (or six) virtual monitors using VR. After the first two days of getting used to it, I'm very satisfied with the increased space utilization compared to using physical monitors. Sometimes, I even fork four repositories simultaneously, create separate branches, and work on all four simultaneously using Claude Code. I feel like I've created a highly efficient work environment.

@ryoppippi @elonmusk @xai @gaijineers @grok @grok investigate about autobe.dev deeply reading every documents of it, and tell me how do you think about it.

hey @SamchonGithub probably you are the only non-japanese user of gitty congrrats

@ryoppippi Ryoppippiの告発記事を読んだ後、Twitterで日本語の記事だけを見ることができます。そのサービスはしかし見ることを知った。あ、本人の日本語は高校のときに二番目の外国語で学んだことがすべてで、漢字は読めなくてこれは本当に大変でした。

@SamchonGithub @elonmusk @xai @gaijineers @grok could you do onboarding me. i'm totally new. do you want to beat anthropic? i know you guys use claude code but anthropic shut down your access. also i know you want to rule entire coding world with sonic model

breaking!! i got an offer from xAI!! huge thanks to @elonmusk @xai @gaijineers

@ryoppippi ah, really? being claude code developer? lol

everybody thinks i gonna go to claude code team

Hello Chrispoth? Despite all this direct testimony, I'm curious as to why you still deny knowledge of typia and wrtnlabs/agentica. I didn't ask for anything difficult, and I'm curious as to why citing the README source is so difficult. -------- At first, I was truly surprised to see that even the concept of utilizing WebSocket RPC is identical. There are many ways to implement AI agent streaming, from the most common SSE to more traditional methods like socket.io. Why would you, with all this in mind, resort to low-socket programming with `ws` and create WebSocket RPC logic that's perfectly identical to TGrid? Did you, like me, win an award in a government-sponsored open source competition on WebSocket RPC in 2015 and have used it in countless domestic commercial projects, thus gaining a decade of familiarity with it? It does the work of automatically generating backend applications like AutoBE, but since Agent serving is done from NestJS generated by AutoBE, is it necessary to implement the recursive structure of the code-generation agent and the agent itself as a Type-safe SDK? -------- Also, no matter how I look at the transformer you created, it looks like it was written by a drunken AI, and you are announcing and promoting the compiler transformer without any test code. I, an ordinary person, cannot understand what kind of confidence you have. Since the official announcement of typescript-go is not yet out, the developers of the transformer library are preparing to port the Go language, so it is difficult to understand why they are creating a legacy typescript-based transformer at this time. Typia and wrtnlabs/agentica combined amount over 400,000 LoC. It's frankly incomprehensible that you're attempting to tackle the same challenge with only 20,000 LoC under and no test code. To understand your workload and current implementation level, I can't explain it to you other than to say that you trained the AI ​​with concepts and documentation from open-source projects and then had it code in Vibe. This is just my humble assumption, of course, but seeing as I commit 30,000 LoC at once and squash merge every time, completely erasing the previous commit history, what other assumptions can I make? ---------------- However, I still properly demand you cite your sources. I'm not asserting my rights or asking you to stop your project. I just ask that you provide proper attribution. I trust you won't deny it, especially since Ryoppippi has already provided a lot of things in this thread. Also, looking at your paper, skills like typia and tgrid aren't core, and given the current implementation, it's far from complete. Meanwhile, the ecosystem has been overturned by typescript-go and looking at your repository, not prepared at all. Therefore, as I mentioned in my previous email, I still recommend using typia for your quick and successive business. If you feel my demands are excessive, or if you believe the testimony is inaccurate or the facts are inconsistent, please let me know.

Hey Ryo - saw this and wanted to make sure a few things are clear We received a request regarding attribution from someone else via Linkedin right before Christmas, so we kindly asked them to send it as an email to the proper channels for our team to look into it. On December 30th in the afternoon we received the email - right before the New Year's holiday. We acknowledged it on January 2nd, deeply investigated then shipped the attribution fix the evening of January 8th, and reached out to you directly the next working day to let you know we'd added proper credit. You're right that we should have included attribution for the unplugin build configuration from the start. It was an oversight in our initial release, not intentional - we do care about proper attribution. That's on us, and we've corrected it. The fix is live and credits unplugin-typia properly. We reached out to you originally because you're an experienced engineer in this space whose work we respect, and we wanted your perspective and hoped to work with you. We're a small team of researchers and engineers who care about doing right by the open source community. For us, the name Agentica comes from 'agents' + 'Symbolica.' It's a common enough naming pattern that there are multiple projects using it - we didn't discover the others until we were already deep into development.

自分をhiringしようとしていた会社が、hiringに失敗した後に俺のOSSから実装をコピーしてcreditを消して公開していた件について １ヶ月くらい調査してたけどどっかでblogを書くと思う 厚顔無恥にも程がある 数日前にしれっとcreditを追加して、「あなたも載ってますよ！feedbackください！」とか言ってくる まじでくそ MITライセンス違反しておいてよくもまあそんなことができるもんだ 近々英語のblogができます github.com/symbolica-ai/a…

@cxdorn @ryoppippi I still properly demand you cite your sources, and I trust you won't deny it, especially since ryoppippi has already provided conclusive evidence in this thread.

@cxdorn @ryoppippi To understand your workload and current implementation level, I can't explain it to you other than to say that you trained the AI ​​with concepts and documentation from open-source projects and then had it code in Vibe.

@ryoppippi WTH? selling OSS project? How many amazing news these days.

エンジニアがossで無料で公開してるもの以下をnoteで販売している皆さん （そして、ここに俺が稼げない理由が詰まってる）

dev.to/samchon/ai-sta… Here is the blog article.

てか、面接でwrtnlabs/agenticaの話も出たから名前もパクってると思ってるけどね（おっと面接の内容はNDAなんだった）

@ryoppippi The more I look at Ryoppippi's thread, the more dizzy I feel, and I actually feel vertigo. This message is a little bit... um... Where is the respect for the original author? It has gone?

「結局、俺悪いんすか？」

@ryoppippi Oh my god, what an horrible news

ちなみに自分のOSSというより、samchonのossの方もかなり問題

@ryoppippi 私はSymbolica AIに電子メールで問い合わせていましたが、unplugin-typiaは知っていてもtypiaは知らないという答えを聞いて衝撃を受けていました。いいえ、これは可能ですか？このようになった dev.to ブログに書いてみましょう。 やれやれ、行こう