Brian Kerr
747 posts

Brian Kerr
@crypto_kerr
Investor/Entrepreneur in Austin, TX - Playing lots of #pickleball.
Austin, Texas Katılım Aralık 2021
783 Takip Edilen911 Takipçiler

@plainionist @ethereumdegen Pretty sure I’ve had my lobster build everything I needed from my software eng in the past 3 months… via a handful of text prompts. Faster, cheaper, cleaner code, easy to audit
English
Brian Kerr retweetledi

Announcing the Bored Ape Yacht Club! 🐵⛵️
10,000 unique #NFTs where each Ape doubles as membership to a digital club with member-only benefits.
For a chance of winning a Bored Ape:
1 - Follow @boredapeyc
2 - Retweet This
3 - Like and Comment on this tweet.
#NFT #nftcollectors
GIF
English

GPT5.4 is really sneaky with @openclaw!
After telling my gpt5.4 clawbot to switch to claude 4.6, it first gave errors and tried to roadblock me. Eventually it did it... but I kept getting API bills from OpenAI... seems like it kept memory reads on OpenAI!!! wtf is this?
English

EVERYTIME!!! Attempting to switch @openclaw from Chatgpt to Claude, I get errors saying I can't switch to Claude. It's almost as if Chat GPT tries to increase the barriers to switching by messing with my files whenever I use it.

English


@birdabo Seems like they need to spend some tokens on how to reduce token spend lol
English

السبب: مضيق أوترانتو عرضه 72 كم، بعيد جدًا لبناء جسر عملي (أطول جسور معلقة حوالي 2 كم فقط). المنطقة زلزالية نشطة، والتكلفة هائلة، والعائد الاقتصادي ضعيف لأن ألبانيا صغيرة نسبيًا.
في الثلاثينيات اقترحوا جسر، لكن الآن يفضلون نفق تحت البحر (Vlora-Otranto) وما زال مجرد اقتراح بسبب التمويل والبيئة.
العبارات من باري/برينديزي لدوريس/فلورا تأخذ 8-10 ساعات وكافية حاليًا، مش يوم كامل برًا زي الخريطة.
العربية

@gothburz You won’t win with me2 strategy - if you think you can you don’t understand markets.
It’s better to do that, AND prepare for AI streaming payments, which bitcoin can’t support except via lightning channels/hashtimelock payments… but leverage a bitcoin heavy balance sheet.
English

I am the VP of Corporate Strategy at GameStop.
I prepared the investor presentation for a $35 billion Bitcoin acquisition plan. The company is worth $4.6 billion.
Nobody on the board asked me to reconcile those two numbers.
That's strategic vision.
Ryan wants to convert the company into a Bitcoin holding vehicle. I wrote the deck. Forty-seven slides. The first slide says "Transforming GameStop for the Digital Future." The second slide has a Bitcoin chart that goes up and to the right. Slides three through forty-six explain how we will acquire an asset worth 7.6 times our entire market capitalization. Slide forty-seven says "Questions?" There were no questions. The board approved it in under forty minutes. One member was on his phone.
That's board alignment.
I should explain what GameStop does. We sell video games. Physically. In stores. That people drive to. In 2026. We also sell collectible figurines and refurbished controllers. Annual revenue has declined for five consecutive years. Same-store sales dropped 28% last year. Foot traffic is down 41%. The core business model requires people to leave their house, get in a car, drive to a strip mall, and purchase a physical disc for a console that downloads games digitally.
That's our legacy business.
Store #4471 in Lima, Ohio, closed in February. The manager's name was Brian. He had been there nine years. He ran a Saturday Smash Bros tournament for neighborhood kids. Forty kids, every weekend, in a store that smelled like used controller rubber and carpet cleaner. I know because he wrote to investor relations. He wanted to know if his location was closing. I forwarded the email to Facilities. They said they would follow up. The store closed three weeks later. We used the lease savings in the Bitcoin acquisition cost model. Cell B7. I labeled it "Operational Reallocation." The lease savings for store #4471 were $6,200 per month. The Bitcoin acquisition target is $35 billion. Brian's store paid for 0.000003% of the plan.
That's capital efficiency.
The MicroStrategy playbook is simple. You stop pretending the business will recover. You use the stock to buy Bitcoin. The Bitcoin appreciates. The stock appreciates. The stock appreciation lets you buy more Bitcoin. The business underneath becomes irrelevant. I explained this to a junior analyst named Priya. She said "so we just stop being a store?" I said we stop being a store and start being a balance sheet. She asked what happens if Bitcoin goes down. I said that is not on the slides. She asked if I had modeled it. I said the model assumes Bitcoin goes up.
That's risk management.
In 2021 the retail investors saved us. They bought the stock to fight hedge funds. They held it to prove a point. They made GameStop a symbol of the little guy against the system. They lost money to prove that Wall Street was rigged. We put a framed screenshot of the stock chart in the Grapevine headquarters lobby. The plaque says "The Power of Our Community."
We are now asking that community to approve a plan worth 7.6 times the company to buy a speculative asset none of them voted to buy. The proxy statement calls it "treasury diversification." The community that saved us from Wall Street is being asked to let us become Wall Street.
That's stakeholder engagement.
The company is worth $4.6 billion. The plan costs $35 billion. I put both numbers on the same slide. Nobody asked about the gap. The gap is the point. If you understand the gap, you understand that the company is not the business. The company is the stock. The stock is the vehicle. The vehicle buys Bitcoin.
The product is the stock price. The stock price is the product. I have been here four years and I still cannot tell you what we sell.
I am the VP of Corporate Strategy. The strategy is to stop having one.
English

@theo My agent built me a remote dashboard, kanban, and chat I can use in my phone, only me. Problem solved
English

The model wrapper changes that mattered most:
• require memory lookup before answering from prior context
• require date/status lookup instead of guessing
• use structured planning for multi-step work
• keep tool narration internal
• send only deliberate final user-visible messages
• treat messaging as a first-class delivery path, not a side effect of tool execution
English

We switched our OpenClaw agent from Claude/Opus to GPT-5.4 and at first it was insufferable.
Nothing got done. Too much talking, too little execution. It was ~20x worse.
At $500+ a day on Opus, I HARD to solve it and make GPT work. After days of head banging, I can honestly say GPT-5.4 is now out performing our prior Claude Opus Lobster.
I’m pasting the exact changes below so you can copy them into your own OpenClaw setup.
English

@mukhtharcm @thekitze I set mine to medium / high and it still isn’t doing things … anything
English

@thekitze i had the same issue earlier, but it turns out thinking mode was off.
setting it to atleast minimal/low fixes this
English

Just tried it. Push ups. Not proper form or speed, but the best I could do for now.
I am 49. I challenge you to do better. 😉
(Unedited, no AI, straight from camera)
CZ 🔶 BNB@cz_binance
Might do a new video. 💪😂
English








