Lokcy

On April 13th, I didn't sleep well, and in the afternoon, I played tennis for two hours.

As a result, when I lay down at night, I couldn't fall asleep, and I felt a tight, painful sensation in my chest, which made me panic a bit.

On April 14th, I rushed to the hospital for some tests: blood draw, routine check-up, ECG, CT scan, from around 4 PM until after 7 PM. After reviewing the results, the doctor said everything looked normal.

But then the doctor asked if I often stayed up late. I admitted that I do, and it's pretty intense; I usually sleep around 5 or 6 AM because I can never seem to fall asleep.

He was very straightforward: this time, the results being normal is a stroke of luck, and it doesn't mean my body can keep handling it; after staying up late, if my mental state is off, I shouldn't be pushing myself to exercise.

Thinking about our industry, many people are definitely overextending themselves:
irregular eating habits, chronic late nights, anxiety, and high stress.

This check-up cost a total of 754, and part of it was covered by health insurance, so I didn't have to pay extra.

Money spent is money spent, the most important thing is to get to the bottom of the issues.

Right now, my biggest takeaway is that market trends, work, and efficiency should always come after health.

Once the body feels off, don't tough it out, and don't scare yourself; get to the hospital for a check-up right away.

Health is the most important thing.

When the market's dull, I can actually see what AI can really do.

Lately, the market has been so boring that I've spent all my time exploring AI.

During this time, I've been tinkering the most with OpenClaw, Codex, Hermes, and Claude Code.

The hype around lobster has waned a bit, and the price of the Mac Mini has dropped slightly. I think that’s a good thing. For many, it’s better to hold off on chasing equipment and first think about what AI is actually useful for.

I’ve mainly been using AI for 4 things lately.

1. Trading and experimenting

Initially, I tried various trading skills: OKX, Binance, GMGN, and Polymarket.

I found it almost impossible to profit on Polymarket; it’s likely my strategy was off. Later, I tried OKX's wallet skill to automatically buy and sell memes, and managed to catch a coin that went up several times, but my position was too small to matter. After that, it was mostly small wins and small losses, nothing significant.

I also subscribed to Binance’s Pro bot for a month and have been testing it with small funds. To be honest, the experience is average, with a bit of lag; I've seen both losses and gains, but I still find it not very user-friendly. Maybe I just haven't fully grasped it yet.

2. AI anime and writing novels

AI anime has been really popular lately, so I tried it out.

I tested Seedance 2.0 for several days, but it barely generated anything; buying a membership was a points system and involved a queue, so it felt like a dead end for now.

Then I tried writing novels. This turned out to be the easiest entry point: find a novel skill, give it a general direction, outline it, and then let AI keep writing, before uploading it to Tomato Novel.

I’m currently working on two books, both just starting, and one of them is already signed. I don’t have high expectations for it to make money; it’s more about fulfilling my long-held dream of writing a novel.

3. Public accounts and X

I saw a news piece saying a couple made 10 million by using AI to write public accounts, which sounded unbelievable. But since I used to write public accounts myself and have an old account with a few thousand followers, I was definitely intrigued.

Recently, I’ve picked up my public account again and am also syncing it on X.

My feeling now is that AI’s biggest value in content isn’t “writing for you,” but helping you produce, revise, format, illustrate, and distribute content more consistently.

Since I already had a foundation for my public account, I have a bit of a feel for it. On X, if you really have the capability and output, the ceiling is pretty high, but the entry barrier isn't low, especially concerning the account itself, which can have a huge impact.

4. Writing scripts and automation

I often use AI to write small scripts. Recently, the airdrop industry feels dead, and I’ve lost interest in interacting.

But for creating batch scripts, repetitive processes, and semi-automated tasks, using AI has become really smooth.

Final conclusion

If I had to sum it up in one sentence:

I'm not using AI to find a shortcut to overnight riches; I’m using it to test trading, create content, write novels, and develop scripts, to see which of these are worth pursuing long-term.

That’s what I’ve been doing with AI lately.

What do you usually use AI for?

Feel free to share any pitfalls you've encountered.

Zhang Xue's motorcycle is making waves, not just for clinching the championship, but because it also sparked significant interest from Zhejiang state-owned capital.

First, let’s break down the investment logic. Public information indicates that in January 2026, Zhejiang Chuangtou led an A-round financing of 90 million yuan for Zhang Xue's motorcycle, with a post-investment valuation of about 1.09 billion yuan. By March 28-29, ZXMOTO 820RR-RS scored two consecutive wins in the WorldSSP category at WSBK in Portugal, leading the second place by 3.685 seconds in the first round. This means that Zhejiang state capital didn’t wait for the victory to ride the hype; they placed their bets before the results came in.

Next, let’s look at the legendary story. Zhang Xue's motorcycle was established in 2024 and officially launched in 2025. In 2025, the company’s total output value was about 750 million yuan, but it also posted a loss of 22.78 million yuan during the same period, with R&D investment reaching 69.58 million yuan, accounting for 9.33%. This isn’t a company with already pretty profits; it’s a Chinese motorcycle enterprise still heavily focused on R&D and investment. Because of this, the subsequent championship victory feels more like a genuine breakthrough.

Finally, consider the recognition from the owner and the team. Cheng Junhua, General Manager of Zhejiang Chuangtou, mentioned in a public interview that the team was initially attracted by Zhang Xue's experience, but ultimately decided to pull the trigger not just because the story was compelling, but because they saw this person understood technology, took responsibility, and could unite the team. What truly gets recognized by investment institutions is never just a boss who can tell a good story, but someone who can inspire the team to hustle together.

Therefore, what’s most worth noting isn’t just that "Zhang Xue won," but why Zhejiang state capital was able to grasp Zhang Xue’s potential earlier than the market. The truly savvy capital isn’t the one that rushes in at the peak of hype, but rather the one that has already placed its bets before most people even realize what's happening.

#InvestmentLogic #ZhejiangStateCapital #Motorcycle #Entrepreneurship #MadeInChina

I see some folks still don't know how to install OpenClaw, and the simplest way is as follows:

1. Download a Codex.
2. Register for a GPT account.
3. Log into Codex using your GPT account.
4. In Codex, just say: help me install OpenClaw.
5. Send this WeChat plugin command to OpenClaw:
npx -y @tencent-weixin/openclaw-weixin-cli@latest install
6. Then scan the QR code with WeChat to integrate WeChat into OpenClaw.

Now, about a relay station CLIProxyAPI:
Authorize with your GPT account, then add authorization with your Google account. The more you add, the better your limits will be; the system will automatically switch to the available limits of your uploaded accounts.

Download links:
Codex: https://openai.com/zh-Hans-CN/codex/get-started/
ChatGPT: https://chatgpt.com
CLIProxyAPI: https://github.com/router-for-me/CLIProxyAPI

Yesterday was packed with info, but what hit me the hardest wasn’t the hype, but the reminder that 'your body will eventually cash in on the lifestyle choices you make.'

After seeing news about sudden deaths, I reminded myself again:
People who stay up late, sit for long periods, and don’t exercise really face high risks.
Even if I start working out now, if my schedule is still chaotic, I can't consider it safe.

Changing habits all at once isn’t realistic, so don’t set overly ambitious goals right away.
Start with what you can manage:
Cut back on staying up late once, walk a bit more, take an extra half-hour to rest.
Change at a natural pace; it’s okay to take it slow, just don’t keep overextending yourself.

Life is ultimately limited.
Instead of constantly pushing through, it’s better to rest well, experience life fully, and make the present a bit more comfortable.

Also, I listened to Jay Chou's new song.
Maybe my expectations were too high, but honestly, I didn’t feel much this time and even thought it wasn’t that great.
Perhaps it’s not the song that has changed, but rather, we have.

The biggest shortcut is actually not taking shortcuts.

In Web3, many folks are out here hunting for the wealth code every day, trying to flip their situation in the shortest time possible.
High leverage, hot altcoins, insider info, and hundred-fold myths all seem like shortcuts.

But if you stick around this space long enough, you'll realize:
The more someone tries to take the fast lane, the harder they tend to fall.

Some have made a killing with high leverage once and then start believing in "going all in";
Only to get wrecked by a flash crash or extreme volatility, wiping out months of profits in an instant.

Then there are those who just keep doing the basics:
Researching airdrops, working on testnets, earning staking rewards, reviewing audits, managing risk.
It may look slow and not so thrilling,
But in the end, they have a better chance of surviving and slowly growing their assets.

In Web3, slow is the new fast.
Because shortcuts are all about making money off emotions,
While the slow and steady approach earns you money from the system.

Winning once by luck doesn’t make you a pro;
Staying alive through the system is where true strength lies.

So in the end,
The biggest shortcut is,
Not taking shortcuts.

#Web3 #Crypto #RiskManagement #InvestmentThinking

The art of speaking,
sometimes it's not about what you say,
it's about how you phrase the same thing, and the vibe changes completely.

For example:

"You like someone else's wife"
can be rephrased as:
"The woman you like ended up being someone else's wife."

"I feel drained as soon as work starts"
can be rephrased as:
"You feel exhausted every day, yet you still show up to grind."

"I want to sleep with you"
can be rephrased as:
"I want to wake up with you."

When the language shifts,
what's stiff becomes nuanced,
what's offensive turns into gentle,
and complaints transform into understanding.

Truly skilled folks,
might not say the most,
but they definitely know how to word things right.

I've been cruising through some posts lately and noticed that "uninstalling Lobster" is becoming a hot topic.

I don't think it's that OpenClaw has failed,
but rather that folks are starting to get real about AI Agents.

A lot of people jumped on Lobster because it looked powerful:
big permissions, able to handle numerous tools, heavy automation capabilities.

But when you actually use it, the issues are pretty straightforward:

1. Permissions are too broad, leaving many feeling uneasy
2. Token consumption is rapid, noticeably affecting costs
3. Everyday scenarios aren't that frequent, making such heavy capabilities unnecessary

To put it simply,
not everyone currently needs a high-permission, heavy-process, token-consuming agent.

If you haven't established a stable workflow yet,
only occasionally researching, writing copy, or doing light automation,
then tools like Lobster are likely just "look strong, feel heavy" for you.

So this wave of "uninstalling Lobster"
isn't really anti-AI,
but rather that people are finally starting to differentiate:

what looks powerful
and what they actually need right now.

I actually think this is a good thing.
It shows the logic behind using AI tools is shifting from "chasing the new and strong" to "making choices based on needs".

#AI #OpenClaw #AIAgent

Recently, I picked up two super useful skills.

1) pua: Don’t let the name fool you; it’s essentially a "reverse meltdown execution rectifier." When AI keeps failing, just repeating the same path, not searching or verifying, and trying to push the problem back to the user, it forces the rhythm back on track: first, check the data, read the source code/documentation, verify assumptions, and switch to fundamentally different solutions until results are delivered. It’s great for debugging, integration, and when research hits a wall.

2) skill-creator: Anthropic's official skill engineering template. The new version doesn’t just teach you to write a SKILL.md; it completes the full iteration loop: creation, optimization, evaluation, benchmarking, variance analysis, and trigger description optimization. The core value is refining "usable skills" into "more stable and easier to trigger correctly skills."

One addresses execution, while the other tackles methodology. The former helps AI avoid slacking, while the latter ensures capabilities can be solidified and iterated. For those who often use agents for work, these two tools work really well together.

#AIAgent #Skills #Automation

Recently came across a long-term memory skill worth keeping an eye on: Claude-Mem.

Its value isn't just in letting the model remember a few extra sentences, but in upgrading AI from a one-off Q&A tool to a sustainable collaborative work assistant.

If you're using AI long-term for content creation, coding, research organization, or knowledge management, capabilities like Claude-Mem will be crucial:
1. Retaining context across sessions, so you don’t have to repeatedly explain project backgrounds.
2. Recording preferences, decisions, and constraints, leading to more consistent outputs.
3. Making collaboration continuous, understanding you better the more it's used.
4. More suited for long-term projects rather than one-off inquiries.

Many people find AI not user-friendly enough; the issue often isn't that the model isn't strong enough, but that it lacks a memory layer.
Whoever connects the model, tools, and memory first will gain a significant edge in their workflow.

If you're already high-frequency trading with AI, I'd categorize Claude-Mem as a foundational skill that deserves priority in your setup.

#AI #AIAgent #ClaudeMem #EfficiencyTool

Today I went to the mall out of boredom and bought some snacks.

I found that several stores I used to know have closed down, and there are noticeably more empty shops than before.

The biggest feeling in the past two years is that many industries still appear to be operating on the surface, but when it comes to the actual offline situation, business is much harder than expected.

Just because the mall has people doesn’t mean the stores are making money.
Even if the stores are still open, it doesn’t mean they can really hold on.

Rent, labor, traffic, and downgraded consumption, each one is squeezing the survival space of physical stores.

Sometimes just taking a walk can give you a sense of the real economic temperature now.
Physical businesses are really hard to run.

#Consumption #PhysicalEconomy #BusinessObservation

Today a friend from the crypto circle told me that the company they work for, which focuses on blockchain infrastructure, is struggling to stay afloat. The team has dwindled to just a few people, and they plan to take compensation and leave. I felt quite emotional after hearing this. During times of market downturn, many so-called foundational projects are often forgotten by the market.

On the other hand, AI is elevating personal efficiency to new heights. Things that were once difficult for an individual to accomplish can now be done quickly with the right idea and execution; AI can really accomplish a lot.

So my current thought is quite simple: while there's still time, I should seriously learn AI and try to create something truly valuable and profitable using it. Let's not wait for the market; let's transform ourselves into productive individuals.

Have you had this feeling lately?

OpenAI has officially incorporated Codex best practices into the developer documentation. My biggest takeaway after reading it is that the truly efficient usage is not treating Codex as a search box, but rather as a configurable, accumulative, and reusable engineering partner.

The 5 most important points to learn:
1. When making requests, clearly articulate the goals, context, constraints, and completion standards together.
2. For complex tasks, plan first, then execute.
3. Write high-frequency rules into AGENTS.md.
4. Continue testing, reviewing, and validating after modifying the code.
5. Turn repetitive processes into skills or automations.

The upper limit of AI is often not determined by the model, but by your workflow.

#AI #Codex #OpenAI #workflow

Has anyone around you really made money with OpenClaw?

I have recently seen many people talking about OpenClaw, and quite a few people say "little lobsters can help you make money, help you work, help you improve efficiency," but to be honest, I want to know the real results.

Who has actually made money?
Is it the people who really use it,
or those who can pretend, adjust, and help others deploy it?

There is a lot of discussion about OpenClaw now,
but there don’t seem to be many people who really show stable results and continuous earnings.

So I really want to hear everyone’s real feedback:
Who around you has really made money with OpenClaw?
How did they make it?

#OpenClaw #AI #Crypto #BinanceSquare

The whole internet is talking about crayfish, but the first batch of people who fell into the trap has actually already appeared.

OpenClaw is indeed very popular; many people jumped in for two reasons:
First, they want to use AI to increase efficiency and make money, and second, they want to get on board with the new narrative early.

But the problems are very real.
The current crayfish is not a tool that is truly suitable for ordinary people to participate in without barriers.

What you see is "AI helps you work,"
but what you often don't see are the costs and risks:
You have to set up the environment, grant high permissions, tinker with configurations, and continuously bear API costs.
If you also hand over your wallet, account, or sensitive permissions to the tool, the risks will be further magnified.

So, the ones who make money first are often not the ones raising crayfish,
but those who know how to set up crayfish, adjust crayfish, or even help others deploy crayfish.

For most ordinary users, it currently feels more like a high-barrier tool,
it's not a money printer that you can profit from with your eyes closed.

If you really want to participate, you should first think clearly about three things:
Can you set it up,
Do you know the boundaries of permissions,
Can you bear the ongoing API costs.

My view is very simple:
At this stage, don't rush into FOMO; first understand, then decide whether to get involved.
When the tools are more mature and the barriers are lower, it won't be too late to participate.

What do you think, does crayfish now feel more like an opportunity or more like a filtering barrier?

#OpenClaw #AI #Crypto #BinanceSquare

This wave of Alpha has actually clarified the market preferences.

OpenAI has absorbed the founder of OpenClaw and continues to support OpenClaw, while Meta has directly acquired Moltbook. After the news broke, Lobster and Molt both quickly appeared on Binance Alpha, and their prices were rapidly driven up.

Now, the rise of many coins is not just based on fundamentals,
but on whether three things overlap at the same time:
endorsement from major companies, narrative popularity, and Alpha traffic entry.

As long as these three things appear simultaneously, funds will chase very aggressively.
So those in the front row can reap big rewards, while those in the back row are most likely to experience high volatility.

This also indicates one thing:
Alpha's strongest ability right now may not just be listing coins,
but in amplifying market sentiment.

Do you think that this wave of Lobster and Molt is the beginning of a new round of Alpha narrative,
or an acceleration phase after the peak?

#BinanceAlpha #Crypto #OpenAI #Meta #BNBChain #Alpha

Binance Alpha has recently added new items: lobster and Freedom of Money.

However, the overall market trend has been weak during this time, and the number of new projects on Alpha has noticeably decreased. More obviously, many projects seem to reach a short-term peak as soon as they hit Alpha, and then quickly fall back, with the window for making money becoming shorter and shorter.

So I'm currently thinking about a question:
Is the benefit of Alpha itself decreasing,
or is the market so bad that most new projects can't hold on?

Of course, those who can truly hold on in the early stages have still made substantial gains.
The problem is that in the current environment, the number of people who can truly hold on and make the right choices is decreasing.

Are you still continuously checking Binance Alpha?
How many points do you roughly check in a day?

#BinanceAlpha #Crypto #BNBChain #Alpha

@BitlayerLabs, I've been following this project for a long time
I still hold the previous helmet NFT

I hope to make a strong comeback through this event

#Bitlayer

Take Off