Peak Founder

@AlorPaschal Try to find anybody to help.

What can I do to surpass this 😩😩😩😩😩

@opencode Please add 3.7 to Zen...

DeepSeek V4 Pro is now available in OpenCode Zen

@stevemordue How will you imagine the pools of vibecs?

If SaaS Apps are supposed to be dead, why are a million vibe-coders trying to create SaaS Apps? 🤔

@allen_explains Awsome lector, и вообще подача такая радует.

Skip Netflix tonight. Spend that hour with Joel Peterson’s Stanford lecture on negotiation instead. It breaks down how to communicate, create leverage, and get better outcomes without sounding desperate or pushy. Most people spend years learning this the hard way. You can get the foundation in 60 minutes. Bookmark it and watch it tonight.

@stevemordue Did you mean, anytime, animal expirience will be out of fashion?

@heydyago @dwarkesh_sp Дима, зато для них сделать что-то значимое элементарно.

@dwarkesh_sp On one hand, completely agree. Once we get there it'll be huge. On the other, we already have 8B human-grade intelligence instances, and it's still hella hard to do anything meaningful.

AI that had human-level intelligence would actually be way above human level in capability. Obviously, if we trained a human-level AI, we could just run way more instances of it in parallel. This is a huge advantage. But it's not the only one. Right now, LLMs are much less sample-efficient than humans, but they can process each individual piece of data much more quickly. A model that was a human-level learner would be able to gain the equivalent of thousands of years of education. AIs can also be way more task-oriented than humans. Even a really well-motivated human worker can only operate with ‘human-level’ effort for relatively short spans of time. AGI could do it continuously.

@stevemordue Давайте лучше цену на пиво поднимем.

My hope? Cost of tokens goes up 10x It will reduce all of the chaffy shit being built and flooding the market by morons just because they can

@stevemordue Мне кажется, должен быть установлен принцип: весь код, который продаётся и лицензируется, должен быть переписан на opensource аналоги и вручён сообществу. Вообще весь.

The most surprising finding in the latest AI reports wasn’t that AI is getting better. Everybody knows that. The surprise is that general-purpose AI tools are already catching up to specialized professional AI tools much faster than expected. BCG found that in 2024, specialized AI tools had a big edge in professional-services work. By 2025, that advantage had basically collapsed in several areas, with general-purpose tools performing neck-and-neck on satisfaction, accuracy, and completeness, and even better on “no rework” output. That should make a lot of AI vendors nervous. Because the standard pitch has been, “Generic AI is fine for casual use, but serious business users need our specialized AI.” Maybe. But maybe not as much as they hoped. If your “AI product” is mostly a prompt library, a nicer interface, or a thin workflow wrapper around someone else’s model, that moat may not be a moat. It may be a puddle. The durable advantage is moving somewhere else: proprietary data, customer context, workflow depth, permissions, auditability, governance, and system-of-record integration. Basically, all the boring enterprise stuff that does not look exciting in a demo but actually matters in production. This is the part a lot of companies still miss. The model keeps getting commoditized. The wrapper keeps getting copied. The UI keeps getting flattened. What is harder to copy is knowing the customer, owning the workflow, controlling the trusted data, and being embedded where the work actually happens. So, the uncomfortable question for a lot of AI software companies is not “How good is your AI?” It is “What do you still have when the model gets better?”

@stevemordue @elonmusk Hi, what is the Tesla Motorhome?

Hey @elonmusk, when can we expect the Tesla Motorhome? I’ll pre-order

@karpathy Как вы думаете, Андрей, необходимо ставить тег "produced by LLM" в контенте, который опубликовывается через LLM gen?

Farzapedia, personal wikipedia of Farza, good example following my Wiki LLM tweet. I really like this approach to personalization in a number of ways, compared to "status quo" of an AI that allegedly gets better the more you use it or something: 1. Explicit. The memory artifact is explicit and navigable (the wiki), you can see exactly what the AI does and does not know and you can inspect and manage this artifact, even if you don't do the direct text writing (the LLM does). The knowledge of you is not implicit and unknown, it's explicit and viewable. 2. Yours. Your data is yours, on your local computer, it's not in some particular AI provider's system without the ability to extract it. You're in control of your information. 3. File over app. The memory here is a simple collection of files in universal formats (images, markdown). This means the data is interoperable: you can use a very large collection of tools/CLIs or whatever you want over this information because it's just files. The agents can apply the entire Unix toolkit over them. They can natively read and understand them. Any kind of data can be imported into files as input, and any kind of interface can be used to view them as the output. E.g. you can use Obsidian to view them or vibe code something of your own. Search "File over app" for an article on this philosophy. 4. BYOAI. You can use whatever AI you want to "plug into" this information - Claude, Codex, OpenCode, whatever. You can even think about taking an open source AI and finetuning it on your wiki - in principle, this AI could "know" you in its weights, not just attend over your data. So this approach to personalization puts *you* in full control. The data is yours. In Universal formats. Explicit and inspectable. Use whatever AI you want over it, keep the AI companies on their toes! :) Certainly this is not the simplest way to get an AI to know you - it does require you to manage file directories and so on, but agents also make it quite simple and they can help you a lot. I imagine a number of products might come out to make this all easier, but imo "agent proficiency" is a CORE SKILL of the 21st century. These are extremely powerful tools - they speak English and they do all the computer stuff for you. Try this opportunity to play with one.

The hottest new programming language is English

появилась идея-рассуждение о том, что составляет кайф при помощи другим, больше похожая на дзен-аксиому. скоро-скоро :)