Binance Square
NathanWalk
544 Posts

NathanWalk

3 Following
104 Followers
224 Liked
Posts
·
--
A fun fact worth 2,600 bucks If you don’t handle your personal paid subscription and billing channel, Twitter verification badges can still result in charges. Getting a badge for free and personal paid subscriptions are actually two separate systems. Twitter won’t pause the personal subscription billing just because you put up the badge. What the hell~
A fun fact worth 2,600 bucks

If you don’t handle your personal paid subscription and billing channel, Twitter verification badges can still result in charges.

Getting a badge for free and personal paid subscriptions are actually two separate systems. Twitter won’t pause the personal subscription billing just because you put up the badge.

What the hell~
Branch to new chat: what gets branched off is just the conversation—previous context is retained. The two chats share the same project folder, so changes to files will affect each other. Worktree: based on git, create two working directories from branches. A worktree is built on top of a git repository, and the project must have at least one commit completed. One-sentence summary: branch chats separate the chat history but not the directory; Worktree separates both the project and the chats.
Branch to new chat: what gets branched off is just the conversation—previous context is retained. The two chats share the same project folder, so changes to files will affect each other.

Worktree: based on git, create two working directories from branches. A worktree is built on top of a git repository, and the project must have at least one commit completed.

One-sentence summary: branch chats separate the chat history but not the directory; Worktree separates both the project and the chats.
Codex’s small feature: “Appshots” Appshots can send the app window at the very front of a page to a ChatGPT chat. Appshots is different from screenshots in that the information it captures also includes various text details behind the images—for example, the link corresponding to a button on a webpage. How to use: 1. Place the window at the very front 2. On Mac, press and hold both Command keys on the left and right sides of the keyboard at the same time. On Windows, press and hold both Alt keys. By default, ChatGPT will open a new chat for an Appshot. If you interact with a chat within 60s, it will instead add it to the most recent chat.
Codex’s small feature: “Appshots”

Appshots can send the app window at the very front of a page to a ChatGPT chat.

Appshots is different from screenshots in that the information it captures also includes various text details behind the images—for example, the link corresponding to a button on a webpage.

How to use:

1. Place the window at the very front
2. On Mac, press and hold both Command keys on the left and right sides of the keyboard at the same time. On Windows, press and hold both Alt keys.

By default, ChatGPT will open a new chat for an Appshot. If you interact with a chat within 60s, it will instead add it to the most recent chat.
Hey, we said we’d reset, right? Yesterday I bought tokens worth 300 and they were gone in a day. I worked hard delivering food; I didn’t gamble or tip any female主播. All the money was taken by you. Reset it now—I need to live.
Hey, we said we’d reset, right?

Yesterday I bought tokens worth 300 and they were gone in a day. I worked hard delivering food; I didn’t gamble or tip any female主播. All the money was taken by you.

Reset it now—I need to live.
#AIA1year, I also made a 3D holographic card This time, @DeAgentAI and the authorized Pokemon distributor are doing a limited “chance encounter” event. Note: it’s a collaboration with the authorized Pokemon distributor, not an official Pokemon collaboration. Participation is easy: 1️⃣ Upload a photo of yourself 2️⃣ Add a shark head and a Pikachu/Charizard element to the image—stickers, filters, or hand-drawing are all fine 3️⃣ Generate and download your 3D holographic card 4️⃣ Post on X with #AIA1stAnniversary and @DeAgentAI Event period: September 19–23. Prizes include PSA 10 graded Pokemon cards, plus 10,000 $AIA shared by 50 winners. Card generation entry: #AIA1stAnniversary #AIA1year
#AIA1year, I also made a 3D holographic card

This time, @DeAgentAI and the authorized Pokemon distributor are doing a limited “chance encounter” event.

Note: it’s a collaboration with the authorized Pokemon distributor, not an official Pokemon collaboration.

Participation is easy:

1️⃣ Upload a photo of yourself
2️⃣ Add a shark head and a Pikachu/Charizard element to the image—stickers, filters, or hand-drawing are all fine
3️⃣ Generate and download your 3D holographic card
4️⃣ Post on X with #AIA1stAnniversary and @DeAgentAI

Event period: September 19–23.

Prizes include PSA 10 graded Pokemon cards, plus 10,000 $AIA shared by 50 winners.

Card generation entry:

#AIA1stAnniversary #AIA1year
Two steps are required to launch the mini program 1. Code review 2. ICP filing The filing process will first be reviewed by Tencent before being submitted to the Ministry of Industry and Information Technology (MIIT). I just received a landline call from Shenzhen. They verified the information—mainly the submitter’s identity details, the mini program name, etc. If any parts don’t meet the requirements, they will help you fix them, such as not mentioning the words “customer” in personal application remarks, and mainly refining textual details. After that, they will submit the filing to MIIT. Last night I submitted the code for review and this morning the code review/filing initial review passed. It seems the final bottleneck will likely come during the MIIT review.
Two steps are required to launch the mini program

1. Code review
2. ICP filing

The filing process will first be reviewed by Tencent before being submitted to the Ministry of Industry and Information Technology (MIIT).

I just received a landline call from Shenzhen. They verified the information—mainly the submitter’s identity details, the mini program name, etc. If any parts don’t meet the requirements, they will help you fix them, such as not mentioning the words “customer” in personal application remarks, and mainly refining textual details. After that, they will submit the filing to MIIT.

Last night I submitted the code for review and this morning the code review/filing initial review passed. It seems the final bottleneck will likely come during the MIIT review.
After buying the Codex Token separately, you can't even see the consumption progress?
After buying the Codex Token separately, you can't even see the consumption progress?
The Codex should let every project define its own icon instead of the current one-size-fits-all folder icons. Only then will a project look visually unique. @thsottiaux
The Codex should let every project define its own icon instead of the current one-size-fits-all folder icons.

Only then will a project look visually unique. @thsottiaux
Vibe is a mini program. It takes 3 hours to develop, but the filing for launch takes 3–22 business days. Kinda ridiculous.
Vibe is a mini program. It takes 3 hours to develop, but the filing for launch takes 3–22 business days. Kinda ridiculous.
The way token limits are handled reflects different companies' operational strategies. I previously always thought that GPT temporarily canceling the 200U subscription was meant to get more people using the best model on Earth. But after thinking it through today, the truth is that a 200U subscription could now be used to generate an additional 300U in revenue—still a bit too naive.
The way token limits are handled reflects different companies' operational strategies. I previously always thought that GPT temporarily canceling the 200U subscription was meant to get more people using the best model on Earth.

But after thinking it through today, the truth is that a 200U subscription could now be used to generate an additional 300U in revenue—still a bit too naive.
Although pPOLY new listings were missed, the value of Binance Alpha is still increasing. Each account has a $10,000 U quota.
Although pPOLY new listings were missed, the value of Binance Alpha is still increasing. Each account has a $10,000 U quota.
A sad story: account number 19 failed to be charged and I missed this reset. After the silly payment just now, I still need to wait another 4 days+. If I’d paid two days later, damn it.
A sad story: account number 19 failed to be charged and I missed this reset.

After the silly payment just now, I still need to wait another 4 days+. If I’d paid two days later, damn it.
The two resets I saved up for have both been used up these past two days. Right now I only have 20% of my tokens left, and tonight they’ll be used up. As for my current situation: a 100U subscription lasts me just 1–2 days. Under the circumstances—while I’ve been doing crazy restructuring of my Web2 workflows—there’s theoretically no reason a 200U Pro plan can’t fully cover me. Over the past month or so I’ve been busy with Web2, and I’ve truly felt everyone’s shared consensus: the smarter the AI, the busier people get. The old saying goes that the stronger the ability, the more you end up living on—you definitely have more to deal with. It’s just that now your capability comes from AI enablement. I just saw that I only have 20% left and I felt completely awful. I went to check with Brother Tibo on his Twitter to see when the reset would happen. And there it was. If you don’t do it, you’ll still end up buying it anyway, hard-pressed as you are. AI is the heroin of the new era—once you take a hit, you’ve got it for life. As long as you can keep “getting a hit” every day without it running out—and sometimes you even get so moved you want to cry—I don’t know if you guys have this kind of symptom. 😂
The two resets I saved up for have both been used up these past two days. Right now I only have 20% of my tokens left, and tonight they’ll be used up.

As for my current situation: a 100U subscription lasts me just 1–2 days. Under the circumstances—while I’ve been doing crazy restructuring of my Web2 workflows—there’s theoretically no reason a 200U Pro plan can’t fully cover me.

Over the past month or so I’ve been busy with Web2, and I’ve truly felt everyone’s shared consensus: the smarter the AI, the busier people get. The old saying goes that the stronger the ability, the more you end up living on—you definitely have more to deal with. It’s just that now your capability comes from AI enablement.

I just saw that I only have 20% left and I felt completely awful. I went to check with Brother Tibo on his Twitter to see when the reset would happen.

And there it was.

If you don’t do it, you’ll still end up buying it anyway, hard-pressed as you are. AI is the heroin of the new era—once you take a hit, you’ve got it for life.

As long as you can keep “getting a hit” every day without it running out—and sometimes you even get so moved you want to cry—I don’t know if you guys have this kind of symptom. 😂
Copy the link and open Pinduoduo. Your help may help an old guy fulfill his dream of getting a free Pinduoduo worth 50 yuan. By the way, why does it feel like everyone around me has successfully cashed out, while I’m always just one last strike away—yes, that’s your strike.🙏 😭
Copy the link and open Pinduoduo.

Your help may help an old guy fulfill his dream of getting a free Pinduoduo worth 50 yuan.

By the way, why does it feel like everyone around me has successfully cashed out, while I’m always just one last strike away—yes, that’s your strike.🙏 😭
PDDUS-1.60%
“ #AIA1year DeAgentAI begins its submission to enterprise clients” When I wrote about @DeAgentAI before, I focused more on Judgment and the zkTLS oracle. This time, the more worth paying attention to is another question: Are enterprises really using it, and can the revenue generated by enterprise calls actually return to $AIA? The AI Token Smart Router that officially went live this time is not an AI Agent for ordinary users—it is an entry point for enterprises to call models. Enterprises don’t need to build a separate integration for each model. With a single Router, they can call DeAgentAI’s self-developed models as well as 38 third-party models. It mainly solves three problems: ▪️ Which model should be called this time?
▪️ How to track usage and handle billing/settlement across different models?
▪️ Who provides the final judgment, and how can the basis be traced? During the pilot operation period, dozens of B2B clients have already been connected. Externally, cumulative calls are about 49 million times. These calls don’t come from DeAgentAI’s own applications—they come from real enterprise usage. Calls generate protocol revenue, and part of that will be used to repurchase $AIA. Enterprise calls → protocol revenue → repurchase $AIA Web3 projects are never short of products and narratives. What they truly lack are customers, revenue, and the relationship between revenue and tokens. From what we can see, Smart Router has already connected these three things. #AIA1stAnniversary #AIA1year
“ #AIA1year
DeAgentAI begins its submission to enterprise clients”

When I wrote about @DeAgentAI before, I focused more on Judgment and the zkTLS oracle.

This time, the more worth paying attention to is another question:

Are enterprises really using it, and can the revenue generated by enterprise calls actually return to $AIA ?

The AI Token Smart Router that officially went live this time is not an AI Agent for ordinary users—it is an entry point for enterprises to call models.

Enterprises don’t need to build a separate integration for each model. With a single Router, they can call DeAgentAI’s self-developed models as well as 38 third-party models.

It mainly solves three problems:

▪️ Which model should be called this time?
▪️ How to track usage and handle billing/settlement across different models?
▪️ Who provides the final judgment, and how can the basis be traced?

During the pilot operation period, dozens of B2B clients have already been connected. Externally, cumulative calls are about 49 million times. These calls don’t come from DeAgentAI’s own applications—they come from real enterprise usage.

Calls generate protocol revenue, and part of that will be used to repurchase $AIA .

Enterprise calls → protocol revenue → repurchase $AIA

Web3 projects are never short of products and narratives. What they truly lack are customers, revenue, and the relationship between revenue and tokens. From what we can see, Smart Router has already connected these three things.

#AIA1stAnniversary #AIA1year
Thank you, @jpmorgan, for the Mid-Autumn Festival goodies. But even though you sent them, I still have to say a few things—you really made the rubber water kettle too eager to stick to hair and lint. I have cats and dogs at home, so after spending half a day cleaning just to take a photo, it was still full of dust. Next time, pay attention. 😆
Thank you, @jpmorgan, for the Mid-Autumn Festival goodies.

But even though you sent them, I still have to say a few things—you really made the rubber water kettle too eager to stick to hair and lint. I have cats and dogs at home, so after spending half a day cleaning just to take a photo, it was still full of dust.

Next time, pay attention. 😆
This HOTO little monkey electric soldering iron is harming people—on Xiaohongshu I’ve seen so many bloggers wildly recommending it. I bought it at a 200% markup for the feel and the craftsmanship, but the quality completely doesn’t live up to the price.
This HOTO little monkey electric soldering iron is harming people—on Xiaohongshu I’ve seen so many bloggers wildly recommending it. I bought it at a 200% markup for the feel and the craftsmanship, but the quality completely doesn’t live up to the price.
With AI, anyone can be a full-stack engineer—able to handle both front-end and back-end, as well as hard and software. An amateur-ish embedded developer can, in just a week, take down something that a whole team of three to five people would have needed three to five months to build. 🤣
With AI, anyone can be a full-stack engineer—able to handle both front-end and back-end, as well as hard and software.

An amateur-ish embedded developer can, in just a week, take down something that a whole team of three to five people would have needed three to five months to build. 🤣
Hardware that used to last a lifetime by relying on experience must also be brought to parity with AI. Even if the principle diagrams drawn by AI today are flawed, PCB layout and routing can still make simple, low-level mistakes. What hardware development calls “experience” is mostly pattern-based things, which is exactly what AI is best at. What we see now—if it doesn’t perform exceptionally well—is largely because large models haven’t been trained with focus, or because the market for this is still relatively niche for the general public.
Hardware that used to last a lifetime by relying on experience must also be brought to parity with AI. Even if the principle diagrams drawn by AI today are flawed, PCB layout and routing can still make simple, low-level mistakes.

What hardware development calls “experience” is mostly pattern-based things, which is exactly what AI is best at. What we see now—if it doesn’t perform exceptionally well—is largely because large models haven’t been trained with focus, or because the market for this is still relatively niche for the general public.
This is what the backend looks like for the 10 days you don’t post on X. Every day, there are a little over 4,000 views from the bottom-level traffic to support it. Then you start posting—your tweet views are only around 400, just like you didn’t post. 🥲
This is what the backend looks like for the 10 days you don’t post on X. Every day, there are a little over 4,000 views from the bottom-level traffic to support it.

Then you start posting—your tweet views are only around 400, just like you didn’t post. 🥲
Log in to explore more content
Join global crypto users on Binance Square
⚡️ Get latest and useful information about crypto.
💬 Trusted by the world’s largest crypto exchange.
👍 Discover real insights from verified creators.
Email / Phone number
Sitemap
Cookie Preferences
Platform T&Cs