Looking at the team's output now, the good parts don't matter whether they were written by you or AI. In fact, being able to work with AI to write good things is the core skill. On the contrary, if a person can't write well even when working with AI, it is basically equivalent to not being able to be taught.

During the interview, Zhang Xiaojun asked Yang Zhilin a question: Your ultimate goal is to develop a general-purpose large-scale model rather than a coding model, right? This question further reflects the fact that the boundaries between how users use, migrate, and form habits for a new AI product, including all large-scale models, agents, and other product forms, are actually quite blurred. Often, exceptional performance in a specific vertical attracts users, while generalization/universality reaching 60% keeps them engaged. This is particularly evident when I use Perplexity and GPT: Perplexity outperforms ChatGPT in market research capabilities (speed); and its practical performance in Google Mail, such as searching and writing, is better than Google's own Gemini. It's precisely because Perplexity outperforms other general-purpose large-scale models in these occasional comparative tasks that I've used it more and more. The same is true for other AI products: The discussion of model vs. no model seems a bit outdated to me, or at least, it's not something front-line builders are considering. The key considerations must be the entry point and the stickiness point. Small problems and small scenarios can be solved one by one, surrounding the city from the countryside, and perplexity can be more useful than GPT and Gemini.

Retweet @m1a0acc: ThumbMaker gets a major upgrade—the latest Banana model delivers even higher click-through rates! Two new professional modes: Portrait Mode: Dat…

I've noticed that many of my college classmates and mutual friends are starting to memorize KPIs in the AI industry. It seems like AI companies from two or three years ago have now shifted to a focus on delivering business results. As a result, I've been receiving a lot of inquiries about AI GTM. On the one hand, I feel like I'm just tuning in. When I was working on web3, no one knew what I was doing. I made it by myself, writing on Twitter. Now, I'm finally seeing some peers coming to the fore. On the other hand, I feel like the field is becoming increasingly crowded, with many up-and-coming talent. But the tides of change wash away the sand, and others haven't fought the battles you've fought. Being unique is your greatest competitive advantage. After three model releases and countless AI product launches in the past two years, I'm only now realizing just how big an edge it is.

I finally understand why some people write books. It's because when you teach a difficult thing to different people more than 10 times, breaking it down to just the steps, and then you find that the other person has learned 70% to 80% of it and leaves, you start wanting to write a book.

I spend an average of $500 per month on AI products. Here's a look at the AI products I abandoned after paying for them: - Claude (impressive writing skills for a short time, replaced by GPT after intensive use) - Deepseek (same as above) - Manus (Deep Research) - Cursor (Composer era) - Lovable (app development) - Macaron (same as above) - Readwise (note-taking product, didn't know how to handle a bunch of excerpted webpage copy, so I stopped paying) Products I've paid for and will continue to use: - ChatGPT - Perplexity (Comet) - Quill (I'm paying but have a low dependency, more for backup and habit) Although many of my subscriptions ended up canceling, I recommend that everyone top up their subscriptions if they can to try out AI products. This isn't to encourage reckless spending, but rather: 1. You get what you pay for. Only through in-depth experience can you truly understand the product's limits: the fastest response speed, the best reasoning and answers, and the most in-depth and advanced interaction methods. In free versions, companies often skimp on costs and performance in hidden ways, leading you to believe "AI" is clumsy, slow, and distrustful (a toxic long-term outcome that's detrimental to building a positive human-machine relationship). 2. Product developers will gain personal insights into when they become dependent, where their payout points are (app-based products generate the most payouts and disappointment), and when they abandon their product. This can provide valuable insights for your own product development, such as using app completeness after three rounds of conversation as a North Star metric. 3. Buying AI is essentially buying privilege. I've seen engineers on our team writing code simultaneously with six cursors, and I've heard from startup friends who've borrowed 100 graphics cards to run coding agents. If I don't write code, I can't even use it. People working in the agency industry envy their ability to deeply tap into AI's intellectual resources and maximize their potential. This has led me to waste a lot of money on apps and web products, as I've been looking for AI products that offer high returns and leverage, and the one I've spent the most on is Perplexity's Comet. AI is a productivity revolution, and these costs are actually insignificant in the pursuit of big results. I've met people who believe in nothing but AI; I've also met people who earn high annual incomes but wait daily for free GPT credits. The gap between the new generation and the old generation is widening. If you can afford it, don't skimp on AI.

It’s hard to be grayscaled yet. I hope it can be grayscaled to gpt5 soon.

I'm not getting enough points for Manus Basic monthly. Should I upgrade to Manus Plus or just buy GPT Pro? I'm looking for deep research and high-density online search capabilities. Anyone who's used it can share their thoughts?

How can I block AI zombie accounts from replying to Twitter posts? I'm really impressed.

Lately, I've realized I spend almost as much time using ChatGPT on my phone as I do working tools like Slack and Google Meet. Add to that the time I spend on my computer and doing deep research, and ChatGPT has become as deeply ingrained in my life as WeChat and Twitter. It's no longer just a work tool; it's playing an increasingly important role in improving my life. So today, I want to share three personal experiences that explain why I believe AI is the best tool for self-discovery and how we should guide it with a fundamental logic to better serve us. https://t.co/BzOlWM4ER4