Brady Long
RT @thisguyknowsai: Startups mostly fail from lack of distribution and no PMF.
And it’s hard to have either without knowing something super well.
So as AI saturates markets it will be hard to build a company w/o having experience in that space.
Damn near impossible actually
tweet
RT @thisguyknowsai: Startups mostly fail from lack of distribution and no PMF.
And it’s hard to have either without knowing something super well.
So as AI saturates markets it will be hard to build a company w/o having experience in that space.
Damn near impossible actually
tweet
Offshore
Video
God of Prompt
RT @alex_prompter: Seedance 2.0 tips that actually matter:
→ Adding "cinematic" or "4K" at the end consistently improves output quality
→ "Slow motion" = smoother, more controlled movement
→ First 20-30 words carry the most weight. Put your subject there.
→ For image-to-video: describe MOTION and CAMERA only. Don't re-describe the image.
→ Start with 5-10 second clips. Lock one shot, then expand.
tweet
RT @alex_prompter: Seedance 2.0 tips that actually matter:
→ Adding "cinematic" or "4K" at the end consistently improves output quality
→ "Slow motion" = smoother, more controlled movement
→ First 20-30 words carry the most weight. Put your subject there.
→ For image-to-video: describe MOTION and CAMERA only. Don't re-describe the image.
→ Start with 5-10 second clips. Lock one shot, then expand.
indeed, Seedance 2.0 is crazy
this is literally my first test https://t.co/2ODCp9rwgl tweet
God of Prompt
The list is directionally right. The framing is backwards.
These companies aren’t dying because “AI replaces SaaS.” They’re dying because the value they captured was never defensible.
Grammarly didn’t own writing. It owned the fact that most people couldn’t prompt a model to edit well. Calendly didn’t own scheduling. It owned the friction between two inboxes. Canva didn’t own design. It owned the gap between “I need a graphic” and “I don’t know Photoshop.”
Every tool on this list is a bridge over a gap that’s closing.
But the take everyone keeps missing: the companies labeled “fucked” aren’t all equally dead. The ones sitting on proprietary data and workflow lock-in (Notion, Intuit, Airtable) have 2-3 years to pivot into AI-native platforms. Some of them will.
The ones that are truly finished are the thin wrappers. The ones where the entire product is “we call an API and add a UI.” That’s Jasper. That’s Writer. That’s half the AI startups nobody put on this list because they haven’t launched yet.
The real extinction event isn’t AI replacing SaaS.
It’s personal agents making the entire category of “simple software that does one thing” irrelevant.
The SaaS graveyard won’t be dramatic. It’ll be slow. People just stop renewing. One by one. Quietly.
tweet
The list is directionally right. The framing is backwards.
These companies aren’t dying because “AI replaces SaaS.” They’re dying because the value they captured was never defensible.
Grammarly didn’t own writing. It owned the fact that most people couldn’t prompt a model to edit well. Calendly didn’t own scheduling. It owned the friction between two inboxes. Canva didn’t own design. It owned the gap between “I need a graphic” and “I don’t know Photoshop.”
Every tool on this list is a bridge over a gap that’s closing.
But the take everyone keeps missing: the companies labeled “fucked” aren’t all equally dead. The ones sitting on proprietary data and workflow lock-in (Notion, Intuit, Airtable) have 2-3 years to pivot into AI-native platforms. Some of them will.
The ones that are truly finished are the thin wrappers. The ones where the entire product is “we call an API and add a UI.” That’s Jasper. That’s Writer. That’s half the AI startups nobody put on this list because they haven’t launched yet.
The real extinction event isn’t AI replacing SaaS.
It’s personal agents making the entire category of “simple software that does one thing” irrelevant.
The SaaS graveyard won’t be dramatic. It’ll be slow. People just stop renewing. One by one. Quietly.
gigafucked:
- grammarly
- calendly
- miro
- retool
- webflow
- langchain
- writer
- harvey
- glean
- expedia
- monday
fucked:
- accenture
- intuit
- notion
- jasper
- canva
- alphasense
- postman
- airtable
- talkdesk
- sierra
- zapier
- replit
- solace
probably fucked:
- cursor
- pilot
- clay
- mercor
naively seems fucked but so competent / plugged in they seem to be figuring it out on the fly anyway:
- linear tweet
X (formerly Twitter)
Tenobrus (@tenobrus) on X
gigafucked:
- grammarly
- calendly
- miro
- retool
- webflow
- langchain
- writer
- harvey
- glean
- expedia
- monday
fucked:
- accenture
- intuit
- notion
- jasper
- canva
- alphasense
- postman
- airtable
- talkdesk
- sierra
- zapier
- replit
- solace…
- grammarly
- calendly
- miro
- retool
- webflow
- langchain
- writer
- harvey
- glean
- expedia
- monday
fucked:
- accenture
- intuit
- notion
- jasper
- canva
- alphasense
- postman
- airtable
- talkdesk
- sierra
- zapier
- replit
- solace…
Jukan
>> Taiwan, US, and Japan Launch Joint Offensive in AI Memory… PSMC Takes on Prototyping and Manufacturing (Economic Daily News)
- As demand for AI technology continues to expand, significant opportunities are emerging in the memory market. There is a growing consensus that traditional memory design and manufacturing frameworks struggle to meet the bandwidth and efficiency requirements of AI systems, driving active development of next-generation memory technologies. Accordingly, PSMC has partnered with Intel (US) and SoftBank Group (Japan) to develop new memory technologies for AI and high-performance computing (HPC).
- This collaboration brings together the industrial capabilities of three countries: Japan's SoftBank is consolidating design and intellectual property through a company called Saimemory; Intel is providing technical support related to memory architecture and stacking technologies; and Taiwan's PSMC along with Japan's Shinko Electric are responsible for prototyping and manufacturing, with a target to complete prototype development by 2027 and achieve commercial production by 2029.
- Industry observers expect this partnership could disrupt the existing memory market structure dominated by SK hynix, Micron, and Samsung, and contribute to establishing new AI memory standards.
$INTC $MU
tweet
>> Taiwan, US, and Japan Launch Joint Offensive in AI Memory… PSMC Takes on Prototyping and Manufacturing (Economic Daily News)
- As demand for AI technology continues to expand, significant opportunities are emerging in the memory market. There is a growing consensus that traditional memory design and manufacturing frameworks struggle to meet the bandwidth and efficiency requirements of AI systems, driving active development of next-generation memory technologies. Accordingly, PSMC has partnered with Intel (US) and SoftBank Group (Japan) to develop new memory technologies for AI and high-performance computing (HPC).
- This collaboration brings together the industrial capabilities of three countries: Japan's SoftBank is consolidating design and intellectual property through a company called Saimemory; Intel is providing technical support related to memory architecture and stacking technologies; and Taiwan's PSMC along with Japan's Shinko Electric are responsible for prototyping and manufacturing, with a target to complete prototype development by 2027 and achieve commercial production by 2029.
- Industry observers expect this partnership could disrupt the existing memory market structure dominated by SK hynix, Micron, and Samsung, and contribute to establishing new AI memory standards.
$INTC $MU
tweet
Jukan
>> China's Glass Fiber Cloth Industry Gears Up for Another Round of Price Hikes — Monthly Adjustments of 10%–15%, Prices Could Double by Year-End (Economic Daily News)
- As glass fiber cloth supply in China continues to tighten, manufacturers are preparing a new round of price increases at 10%–15% on a monthly basis. Cumulative price increases since 2025 have already exceeded 50%, creating a recurring cycle of supply shortages and rising raw material costs.
- Glass fiber cloth is a key raw material in the production of copper-clad laminates (CCL), and the price hikes are expected to exert upward cost pressure across the entire PCB industry supply chain. Industry participants assess that if these supply constraints persist, prices could double from current levels by the end of 2026.
tweet
>> China's Glass Fiber Cloth Industry Gears Up for Another Round of Price Hikes — Monthly Adjustments of 10%–15%, Prices Could Double by Year-End (Economic Daily News)
- As glass fiber cloth supply in China continues to tighten, manufacturers are preparing a new round of price increases at 10%–15% on a monthly basis. Cumulative price increases since 2025 have already exceeded 50%, creating a recurring cycle of supply shortages and rising raw material costs.
- Glass fiber cloth is a key raw material in the production of copper-clad laminates (CCL), and the price hikes are expected to exert upward cost pressure across the entire PCB industry supply chain. Industry participants assess that if these supply constraints persist, prices could double from current levels by the end of 2026.
tweet
Offshore
Photo
Moon Dev
lifetime all access
lifetime all access was a very bad trade for me
people are literally paying $297/mo for it
at this price for lifetime access, that is a massive discount
so i have to remove it, if you want in, get in: https://t.co/FKIm9UgUXx
moon dev https://t.co/6zyqrqfZLo
tweet
lifetime all access
lifetime all access was a very bad trade for me
people are literally paying $297/mo for it
at this price for lifetime access, that is a massive discount
so i have to remove it, if you want in, get in: https://t.co/FKIm9UgUXx
moon dev https://t.co/6zyqrqfZLo
tweet
Offshore
Video
Moon Dev
Your been getting scammed and don’t even realize it
You don’t need a Mac Pro or some expensive hardware
I cracked the code so I’m leaving. https://t.co/Grr4UU373w
tweet
Your been getting scammed and don’t even realize it
You don’t need a Mac Pro or some expensive hardware
I cracked the code so I’m leaving. https://t.co/Grr4UU373w
tweet
Offshore
Photo
Chips & SaaS
RT @fiscal_ai: Google is the most profitable company in the world.
Can any company surpass them in the next 10 years?
$GOOGL $AAPL $MSFT $AMZN $NVDA https://t.co/pCqL4TwXWw
tweet
RT @fiscal_ai: Google is the most profitable company in the world.
Can any company surpass them in the next 10 years?
$GOOGL $AAPL $MSFT $AMZN $NVDA https://t.co/pCqL4TwXWw
tweet
Offshore
Video
Moon Dev
Chasing Simons: How I Used DeepSeek R1 to Build an Autonomous 24/7 Quant Fleet
most traders are getting wrecked by high frequency firms while a new open source model just handed us the keys to the kingdom. if you think you need a computer science degree to build a profitable system you are already losing a war you haven't even joined yet. there is a specific reason why deepseek r1 is scaring the legacy ai giants and it has everything to do with how it thinks through complex trading logic
i spent hundreds of thousands of dollars on developers in the past because i was convinced i couldn't handle the syntax myself. that mistake cost me more than just money it cost me the ability to iterate at the speed of the market. when you realize that code is the great equalizer you start to see that liquidations are just a symptom of emotional trading
the entry of deepseek r1 into the space changes the math for every solo trader. it uses a reasoning process that doesn't just guess the next word but actually builds a mental model of the code before it starts typing. i started using it to build out a funding rate bot because i wanted a system that could earn while i slept without the stress of directional guessing
most people look at funding rates as a boring small percentage play but they miss the compounding power of a bot that never gets tired. r1 was able to handle the complex api calls for hyperliquid with a precision that usually requires hours of debugging. the trick is in how you prompt the model to think before it executes the python script
if you just ask for a bot it will give you a template but if you ask it to reason through the risk management first you get a weapon. i found that giving the ai a clear role as a quantitative engineer makes it prioritize the edge cases that usually blow up accounts. this is how we move from gambling on candle sticks to running a serious business
the infrastructure is the part that usually trips people up especially when they try to connect to an exchange for the first time. hyperliquid has become my go to because the transparency is unmatched and the api is built for speed. r1 helped me bridge the gap between a raw idea and a script that actually places orders in the order book
one of the biggest loops to close is the fear of losing your entire balance due to a coding error. i mitigate this by having the ai write its own test cases before we ever go live with a single dollar. seeing the bot pass a simulation built by a reasoning model gives you the confidence to finally turn off the manual terminal
the real magic happened when i asked the model to optimize my social arbitrage strategy. it recognized patterns in how news moves small cap tokens that i had overlooked for years. it turns out that speed is only half the battle the other half is knowing exactly which data points to ignore
iterating to success means you are going to see errors and that is where most people quit and go back to over trading. with r1 you just feed the error back into the reasoning loop and it explains exactly why the logic failed. it feels like having a senior developer sitting next to you who never gets frustrated and only costs pennies to run
i remember the days when i would stay up until four am staring at forty x leverage positions and praying for a bounce. code changed that by removing the man from the middle of the trade and replacing him with a cold logic. now i spend my time building more bots instead of hoping the market does what i want it to do
deepseek is especially good at handling the python libraries like ccxt or the specific sdks for decentralized exchanges. it understood the difference between a market order and a limit order in a way that prevented unnecessary slippage. every basis point you save in execution adds up to a massive difference at the end of the year
the scariest part for the big players is that this technology is now accessible to anyone with a laptop and an internet connection. the moat t[...]
Chasing Simons: How I Used DeepSeek R1 to Build an Autonomous 24/7 Quant Fleet
most traders are getting wrecked by high frequency firms while a new open source model just handed us the keys to the kingdom. if you think you need a computer science degree to build a profitable system you are already losing a war you haven't even joined yet. there is a specific reason why deepseek r1 is scaring the legacy ai giants and it has everything to do with how it thinks through complex trading logic
i spent hundreds of thousands of dollars on developers in the past because i was convinced i couldn't handle the syntax myself. that mistake cost me more than just money it cost me the ability to iterate at the speed of the market. when you realize that code is the great equalizer you start to see that liquidations are just a symptom of emotional trading
the entry of deepseek r1 into the space changes the math for every solo trader. it uses a reasoning process that doesn't just guess the next word but actually builds a mental model of the code before it starts typing. i started using it to build out a funding rate bot because i wanted a system that could earn while i slept without the stress of directional guessing
most people look at funding rates as a boring small percentage play but they miss the compounding power of a bot that never gets tired. r1 was able to handle the complex api calls for hyperliquid with a precision that usually requires hours of debugging. the trick is in how you prompt the model to think before it executes the python script
if you just ask for a bot it will give you a template but if you ask it to reason through the risk management first you get a weapon. i found that giving the ai a clear role as a quantitative engineer makes it prioritize the edge cases that usually blow up accounts. this is how we move from gambling on candle sticks to running a serious business
the infrastructure is the part that usually trips people up especially when they try to connect to an exchange for the first time. hyperliquid has become my go to because the transparency is unmatched and the api is built for speed. r1 helped me bridge the gap between a raw idea and a script that actually places orders in the order book
one of the biggest loops to close is the fear of losing your entire balance due to a coding error. i mitigate this by having the ai write its own test cases before we ever go live with a single dollar. seeing the bot pass a simulation built by a reasoning model gives you the confidence to finally turn off the manual terminal
the real magic happened when i asked the model to optimize my social arbitrage strategy. it recognized patterns in how news moves small cap tokens that i had overlooked for years. it turns out that speed is only half the battle the other half is knowing exactly which data points to ignore
iterating to success means you are going to see errors and that is where most people quit and go back to over trading. with r1 you just feed the error back into the reasoning loop and it explains exactly why the logic failed. it feels like having a senior developer sitting next to you who never gets frustrated and only costs pennies to run
i remember the days when i would stay up until four am staring at forty x leverage positions and praying for a bounce. code changed that by removing the man from the middle of the trade and replacing him with a cold logic. now i spend my time building more bots instead of hoping the market does what i want it to do
deepseek is especially good at handling the python libraries like ccxt or the specific sdks for decentralized exchanges. it understood the difference between a market order and a limit order in a way that prevented unnecessary slippage. every basis point you save in execution adds up to a massive difference at the end of the year
the scariest part for the big players is that this technology is now accessible to anyone with a laptop and an internet connection. the moat t[...]
Offshore
Moon Dev Chasing Simons: How I Used DeepSeek R1 to Build an Autonomous 24/7 Quant Fleet most traders are getting wrecked by high frequency firms while a new open source model just handed us the keys to the kingdom. if you think you need a computer science…
hat the hedge funds built around their proprietary systems is evaporating. we are now in an era where the best ideas win not just the ones with the most funding
a lot of people ask me if they should start with gpt four or claude but i tell them to look at the reasoning tokens of r1. there is a depth to the output that allows it to catch logical fallacies in your trading plan before they cost you real money. it is about building a system that is robust enough to survive a black swan event
my journey from getting liquidated to running a fully automated system was paved with a lot of mistakes and expensive lessons. i share all of this because i want you to realize that you are closer to your goals than you think. the bot we built today is just the beginning of a larger shift in how we approach wealth
when the market is volatile most traders panic and make the worst decisions of their careers. my bots just see another data point and execute the risk controls that were programmed into them months ago. that level of peace of mind is the real reason why i advocate for automation every single day
we are moving into a phase where your ability to communicate with ai will dictate your financial trajectory. it is no longer about how well you can click a buy button but how well you can architect a system. r1 is just the first of many tools that will make this process even faster
if you keep waiting for the perfect time to start you are just giving the institutions more time to take your money. the barrier to entry is gone and the tools are literally sitting on your screen right now. it is time to stop being the liquidity for someone else and start being the one who owns the systems
i will keep iterating and building in the open because that is how we all get better. the bots are trading right now while i talk to you and that is the power of code. i hope you take this and start building your own version of an automated future
tweet
a lot of people ask me if they should start with gpt four or claude but i tell them to look at the reasoning tokens of r1. there is a depth to the output that allows it to catch logical fallacies in your trading plan before they cost you real money. it is about building a system that is robust enough to survive a black swan event
my journey from getting liquidated to running a fully automated system was paved with a lot of mistakes and expensive lessons. i share all of this because i want you to realize that you are closer to your goals than you think. the bot we built today is just the beginning of a larger shift in how we approach wealth
when the market is volatile most traders panic and make the worst decisions of their careers. my bots just see another data point and execute the risk controls that were programmed into them months ago. that level of peace of mind is the real reason why i advocate for automation every single day
we are moving into a phase where your ability to communicate with ai will dictate your financial trajectory. it is no longer about how well you can click a buy button but how well you can architect a system. r1 is just the first of many tools that will make this process even faster
if you keep waiting for the perfect time to start you are just giving the institutions more time to take your money. the barrier to entry is gone and the tools are literally sitting on your screen right now. it is time to stop being the liquidity for someone else and start being the one who owns the systems
i will keep iterating and building in the open because that is how we all get better. the bots are trading right now while i talk to you and that is the power of code. i hope you take this and start building your own version of an automated future
tweet
Offshore
Video
Moon Dev
I wasn’t going to post this and say goodbye bc of openclaw
But I’ve decided to give you the heads up
On Thursday I will be gone, heads down, I found something huge. https://t.co/FjgsOaYPFf
tweet
I wasn’t going to post this and say goodbye bc of openclaw
But I’ve decided to give you the heads up
On Thursday I will be gone, heads down, I found something huge. https://t.co/FjgsOaYPFf
tweet