Evan

Tech & Art

[Claude @claudeai]

Kay Zhu is the co-founder and CTO of @genspark_ai, the all-in-one AI workspace built on Claude.

In a market moving this fast, where anyone can build, he thinks the team is what makes the difference:

[Evan @Evan3point]

@yanhua1010 Helio is genuinely useful, in 5 minutes, I built an info tracker I’d been wanting for a while: it monitors AI news based on my needs, then structures and enriches it the way I want. Still a few bugs at this stage, but very promising.

[Yanhua @yanhua1010]

https://t.co/qIWuePdel2

[Evan @Evan3point]

@yanhua1010 Helio is genuinely useful, in 5 minutes, I built an info tracker I’d been wanting for a while: it monitors AI news based on my needs, then structures and enriches it the way I want. Still a few bugs at this stage, but very promising.

[Yanhua @yanhua1010]

突然发现之前折腾什么OpenClaw，Hermes全他妈浪费时间。

这才是真正的AI Native，目前体验下来最舒服的产品

[Helio @helioim_ai]

Sheet0 is live🚀

[Wells Wang @wellswfwang]

Excel was built for the '80s.
@sheet0ai is built for today.

Sheet0 is the first Level 4 data agent — a spreadsheet that drives itself.

Made for 100% accuracy, 0 AI hallucinations, and fully explainable results.

Early preview is live — Comment or quote for invite code. 🧵

[Evan @Evan3point]

@lovart_ai I’m a UX designer and blogger based in China with a growing audience. Your product looks fantastic, and I’d love to join the closed beta—could you please send me an invite code? Really appreciate it, thank you!

[Nat Friedman @natfriedman]

The deepseek team is obviously really good. China is full of talented engineers. Every other take is cope. Sorry.

[Andrej Karpathy @karpathy]

DeepSeek (Chinese AI co) making it look easy today with an open weights release of a frontier-grade LLM trained on a joke of a budget (2048 GPUs for 2 months, $6M).

For reference, this level of capability is supposed to require clusters of closer to 16K GPUs, the ones being brought up today are more around 100K GPUs. E.g. Llama 3 405B used 30.8M GPU-hours, while DeepSeek-V3 looks to be a stronger model at only 2.8M GPU-hours (~11X less compute). If the model also passes vibe checks (e.g. LLM arena rankings are ongoing, my few quick tests went well so far) it will be a highly impressive display of research and engineering under resource constraints.

Does this mean you don't need large GPU clusters for frontier LLMs? No but you have to ensure that you're not wasteful with what you have, and this looks like a nice demonstration that there's still a lot to get through with both data and algorithms.

Very nice & detailed tech report too, reading through.

[Tsarathustra @tsarnick]

Perplexity CEO Aravind Srinivas says restrictions on chip imports to China are forcing them to innovate efficient solutions to AI model training, with DeepSeek trained on only 2048 H800 GPUs, making them the equivalent of DOGE for AI

[Marc Andreessen 🇺🇸 @pmarca]

Deepseek R1 is one of the most amazing and impressive breakthroughs I’ve ever seen — and as open source, a profound gift to the world. 🤖🫡

[Tanishq Mathew Abraham, Ph.D. @iScienceLuvr]

DeepSeek CEO's thesis is that in AI, there isn't any moat in being closed-source, but rather in having a talented team that can keep innovating

[Yann LeCun @ylecun]

To people who think
"China is surpassing the US in AI"
the correct thought is
"Open source models are surpassing closed ones"
See ⬇️⬇️⬇️

[Evan @Evan3point]

@reasan_mirasan this is truly great

[Evan @Evan3point]

@blader @genspark_ai fact check agent and data search agent are both outstanding

[Evan @Evan3point]

@Corallo44957336 私は数週間Gensparkを使ってみましたが、かなり便利だと思います。今では伝統的な検索エンジンをほとんど使わなくなりました。直接答えが得られて、とても正確で網羅的です

[Robin Delta @heyrobinai]

I think AI just killed Google search.

Here’s https://t.co/WfktABqmqo.

I’ve spent countless hours sifting through ads and biased content. Genspark gives me direct, high-quality info from AI agents, saving me time and hassle.👇

#Genspark

[Wells Wang @wellswfwang]

Introducing NPi, an open-source platform providing out-of-the-box Tool-use APIs to empower AI agents with the ability to operate and interact with a diverse array of software tools and applications.

For example, you can ask the AI agent to create a calendar event for you with just one line of code: google_calendar.chat("schedule a meeting with Kai on Friday").

With NPi, you can:

- Integrate or expand "Tool Use" capabilities for AI agents.

- Interact with a new form of APIs that can interpret your task description in natural language and execute actions for tools and apps in a rule-based way.

- Handle annoying stuff: State Management, Version Control, Availability, Authorization flow and more!

- Continuously growing platform with our responsive support and a collaborative community.

For the first release (v0.0.1), we support Gmail, Google Calendar, GitHub, Discord, Twitter, and a General Browser Agent.
Talk is cheap, let's show you some demo! (1/n)

#opensource #buildinginpublic #LLM #SanFrancisco #OpenAI #ChatGPT #GitHub #functioncall #AIagents

[Wells Wang @wellswfwang]

@thedigitaldr Actually previous reply was done by NPi😂, here is the video! not any rehearsal and before that, we never test this case🤣

This video was clipped, the original version is ~5mins for some repeated but unnecessary steps.

We may optimize it less than 2mins in next release🧐