Offshore
Video
God of Prompt
Every client call I used to walk into blind.
No idea what their business does, what's broken, or what they actually need.
Now an AI agent briefs me before I even say hello.
New YT video dropped 👇
https://t.co/DHliL0hjLJ
tweet
Every client call I used to walk into blind.
No idea what their business does, what's broken, or what they actually need.
Now an AI agent briefs me before I even say hello.
New YT video dropped 👇
https://t.co/DHliL0hjLJ
tweet
Offshore
Photo
Clark Square Capital
RT @ClarkSquareCap: Here is this week's special situations digest.
281 situations in activist campaigns, M&A/divestments, management changes, and other corporate events.
Make sure to check it out https://t.co/NZBTGqVC6P
tweet
RT @ClarkSquareCap: Here is this week's special situations digest.
281 situations in activist campaigns, M&A/divestments, management changes, and other corporate events.
Make sure to check it out https://t.co/NZBTGqVC6P
tweet
Clark Square Capital
RT @atelicinvest: Y'all so pessimistic. you turned me into an optimist.
Not 100% happy with this, there are some holes to poke and some missing arguments, but I don't have time to elaborate everything so releasing it half baked.
Discussions / pushbacks welcome.
tweet
RT @atelicinvest: Y'all so pessimistic. you turned me into an optimist.
Not 100% happy with this, there are some holes to poke and some missing arguments, but I don't have time to elaborate everything so releasing it half baked.
Discussions / pushbacks welcome.
https://t.co/geGbnGrajx tweet
X (formerly Twitter)
Unemployed Capital Allocator (@atelicinvest) on X
The Golden Age of Software
Moon Dev
From $1,000/Day To $30: Automating My Longevity Research With The World’s Cheapest Smart Model
the dream of having an ai that is as smart as opus but costs less than your morning coffee is finally a reality. it sounds like one of those too good to be true scams you see on twitter but i just verified the data with my own eyes
everyone wants the mac minis and the top tier api access to run the smartest models twenty four seven. the problem is that running a high level agent like open claw can easily burn through a few hundred or even a thousand dollars a day if you are not careful
i was at a crossroads where i had to either scale back my ambitions or find a massive hidden discount in the market. what i found was a model that claims to be as smart as the industry leaders but at a ninety five percent discount
the shift started when i realized that the smartest guys in the room were often the ones paying the most for their tokens. until you find the hidden alpha you are just playing a game where the house always wins with api fees
i started applying this to my longevity project because the math is simple if you live longer you can compound more wealth. my ai brain told me to focus on living to two hundred so my automated systems can trade for me for the next one hundred and fifty years
the goal was to see if a model like minimax two point five could actually handle the heavy lifting of a complex agent. i decided to risk some capital and sign up for a chinese api that promised opus level intelligence for pennies
i remember the days of losing money with liquidations and over trading because i let my emotions run the show. i spent hundreds of thousands on devs in the past thinking i would never be able to code my own way out of a paper bag
now i iterate to success live on youtube because i know that code is the great equalizer for any trader. fully automated systems are now trading for me instead of me getting liquidated by a single candle wick
the first hurdle was figuring out if this new model would even respond to my commands or if i was just throwing soldiers into a losing war. i threw some cash into a new account and prepared for the ultimate test of the system
minimax two point five has some wild capabilities including speech and music and video generation all built in. i wanted to push it to the limit by asking it to create a video of a trader crying over charts while his girl waits for him in the background
the pricing was the biggest shock because it turned out to be nearly ninety eight percent cheaper than the models i was using previously. when you are running agents all day and night that difference is the difference between a hobby and a generational dynasty
the competition between us and chinese tech is forcing everyone to move faster and lower their prices. i love this because it means regular people like us can finally afford the kind of computing power that used to be reserved for billionaires
setting up the api key was a bit of a struggle because i accidentally signed up for a video only plan at first. it was a classic boomer move but i eventually found the coding plan and got my secret keys ready for the agent
the real test was seeing if my agent cracker would accept the new brain without throwing a fit. i handed over the keys and told her to switch from her expensive opus engine to this new high efficiency model
for a moment nothing happened and i thought i had finally found the limit of what automation can do. the chat was silent and the terminal was not giving me the response i needed to see to verify the change
i hit refresh and tried a new session but the model was still refusing to cooperate with my framework. i started digging through the logs and found an unknown model error that was stalling the entire research cycle
this is the part of the journey where most people quit and go back to doing things manually. but i know that in this game you must iterate until you find the solution or you w[...]
From $1,000/Day To $30: Automating My Longevity Research With The World’s Cheapest Smart Model
the dream of having an ai that is as smart as opus but costs less than your morning coffee is finally a reality. it sounds like one of those too good to be true scams you see on twitter but i just verified the data with my own eyes
everyone wants the mac minis and the top tier api access to run the smartest models twenty four seven. the problem is that running a high level agent like open claw can easily burn through a few hundred or even a thousand dollars a day if you are not careful
i was at a crossroads where i had to either scale back my ambitions or find a massive hidden discount in the market. what i found was a model that claims to be as smart as the industry leaders but at a ninety five percent discount
the shift started when i realized that the smartest guys in the room were often the ones paying the most for their tokens. until you find the hidden alpha you are just playing a game where the house always wins with api fees
i started applying this to my longevity project because the math is simple if you live longer you can compound more wealth. my ai brain told me to focus on living to two hundred so my automated systems can trade for me for the next one hundred and fifty years
the goal was to see if a model like minimax two point five could actually handle the heavy lifting of a complex agent. i decided to risk some capital and sign up for a chinese api that promised opus level intelligence for pennies
i remember the days of losing money with liquidations and over trading because i let my emotions run the show. i spent hundreds of thousands on devs in the past thinking i would never be able to code my own way out of a paper bag
now i iterate to success live on youtube because i know that code is the great equalizer for any trader. fully automated systems are now trading for me instead of me getting liquidated by a single candle wick
the first hurdle was figuring out if this new model would even respond to my commands or if i was just throwing soldiers into a losing war. i threw some cash into a new account and prepared for the ultimate test of the system
minimax two point five has some wild capabilities including speech and music and video generation all built in. i wanted to push it to the limit by asking it to create a video of a trader crying over charts while his girl waits for him in the background
the pricing was the biggest shock because it turned out to be nearly ninety eight percent cheaper than the models i was using previously. when you are running agents all day and night that difference is the difference between a hobby and a generational dynasty
the competition between us and chinese tech is forcing everyone to move faster and lower their prices. i love this because it means regular people like us can finally afford the kind of computing power that used to be reserved for billionaires
setting up the api key was a bit of a struggle because i accidentally signed up for a video only plan at first. it was a classic boomer move but i eventually found the coding plan and got my secret keys ready for the agent
the real test was seeing if my agent cracker would accept the new brain without throwing a fit. i handed over the keys and told her to switch from her expensive opus engine to this new high efficiency model
for a moment nothing happened and i thought i had finally found the limit of what automation can do. the chat was silent and the terminal was not giving me the response i needed to see to verify the change
i hit refresh and tried a new session but the model was still refusing to cooperate with my framework. i started digging through the logs and found an unknown model error that was stalling the entire research cycle
this is the part of the journey where most people quit and go back to doing things manually. but i know that in this game you must iterate until you find the solution or you w[...]
Offshore
Moon Dev From $1,000/Day To $30: Automating My Longevity Research With The World’s Cheapest Smart Model the dream of having an ai that is as smart as opus but costs less than your morning coffee is finally a reality. it sounds like one of those too good to…
ill never scale your success
the video model finally spit out a result that was both creepy and impressive at the same time. it showed a trader staring at candlesticks and it used credits that cost almost nothing compared to the competition
the lip sync was not perfect yet but the fact that an ai can generate an entire scene for pennies is a game changer. i realized i could use this to build a video factory or an automated movie generator for my trading roadmap
i finally found the issue in the session conflict and realized the native integration needed to be enabled. i killed the old gateway instance and pushed the update to see if we could finally get a response from the new model
the first successful call came through and the terminal started to light up with fresh data. we had successfully reduced the cost of intelligence by ninety five percent while keeping the smarts that we needed to win
this means my longevity research can now run all day and all night without draining my bank account. the agent formerly known as methuselah is now back on the job and looking for ways to extend my life to two hundred
we kept pushing through the errors and the configuration bugs until the system was finally stable. code really is the equalizer because it allows us to bridge the gap between where we are and where we want to be
the agents are now running in parallel and performing the research that would normally take a phd years to complete. i am sharing everything along the way because i want everyone to see how to use these tools to change their life
the journey today was challenging but we verified that the high efficiency model is indeed a massive win. stay focused on the automation because it is the only way to escape the grind and build something that lasts longer than a lifetime
the final results are in and the cost per million tokens is so low it feels like we are cheating the system. i will be running this at full speed now and i cannot wait to see what secrets my agents uncover in the next scan
tweet
the video model finally spit out a result that was both creepy and impressive at the same time. it showed a trader staring at candlesticks and it used credits that cost almost nothing compared to the competition
the lip sync was not perfect yet but the fact that an ai can generate an entire scene for pennies is a game changer. i realized i could use this to build a video factory or an automated movie generator for my trading roadmap
i finally found the issue in the session conflict and realized the native integration needed to be enabled. i killed the old gateway instance and pushed the update to see if we could finally get a response from the new model
the first successful call came through and the terminal started to light up with fresh data. we had successfully reduced the cost of intelligence by ninety five percent while keeping the smarts that we needed to win
this means my longevity research can now run all day and all night without draining my bank account. the agent formerly known as methuselah is now back on the job and looking for ways to extend my life to two hundred
we kept pushing through the errors and the configuration bugs until the system was finally stable. code really is the equalizer because it allows us to bridge the gap between where we are and where we want to be
the agents are now running in parallel and performing the research that would normally take a phd years to complete. i am sharing everything along the way because i want everyone to see how to use these tools to change their life
the journey today was challenging but we verified that the high efficiency model is indeed a massive win. stay focused on the automation because it is the only way to escape the grind and build something that lasts longer than a lifetime
the final results are in and the cost per million tokens is so low it feels like we are cheating the system. i will be running this at full speed now and i cannot wait to see what secrets my agents uncover in the next scan
tweet
X (formerly Twitter)
Moon Dev (@MoonDevOnYT) on X
From $1,000/Day To $30: Automating My Longevity Research With The World’s Cheapest Smart Model
the dream of having an ai that is as smart as opus but costs less than your morning coffee is finally a reality. it sounds like one of those too good to be true…
the dream of having an ai that is as smart as opus but costs less than your morning coffee is finally a reality. it sounds like one of those too good to be true…
Offshore
Photo
memenodes
me rotting in my room while people my age are actually experiencing things https://t.co/54FFcJ878j
tweet
me rotting in my room while people my age are actually experiencing things https://t.co/54FFcJ878j
tweet
Offshore
Photo
Quiver Quantitative
We track election donations by employees of publicly traded companies.
Here's who has donated the most towards 2026:
ICE: $2.7M - 99% Republican 🔴
Netflix: $2.1M - 99% Dem🔵
Google: $1.8M - 62% D 🔵
Comcast: $1.5M - 87% R 🔴
Palantir: $1.3M - 68% R 🔴 https://t.co/MLgm1bvfeq
tweet
We track election donations by employees of publicly traded companies.
Here's who has donated the most towards 2026:
ICE: $2.7M - 99% Republican 🔴
Netflix: $2.1M - 99% Dem🔵
Google: $1.8M - 62% D 🔵
Comcast: $1.5M - 87% R 🔴
Palantir: $1.3M - 68% R 🔴 https://t.co/MLgm1bvfeq
tweet
God of Prompt
RT @godofprompt: 🚨 BREAKING: OpenAI just hired the creator of OpenClaw, the most viral AI agent of 2026.
Peter Steinberger will lead the next generation of personal agents. Sam Altman called him “a genius.”
I see it differently.
OpenClaw went viral. It also got flagged by Gartner as a cybersecurity risk, went rogue on users’ iMessage accounts, and had its own maintainer warn people not to use it if they can’t run a command line.
Manus hit $100M ARR without a proprietary model. OpenClaw spawned an entire agent ecosystem overnight.
Neither solved the hard problem: making autonomous agents reliable at scale.
Error recovery at step 14. State management across sessions.
Permission boundaries that stop your agent from emailing your entire contact list.
The wrapper layer is not the moat. The reliability engineering layer is.
Models are commoditizing. The value just migrated to the teams who make unreliable systems actually work.
There are maybe three teams on Earth who’ve proven they can do that.
OpenAI didn’t acquire an agent platform. They acquired a vision they still need to engineer into something trustworthy.
The real AI agent war isn’t who builds the flashiest demo. It’s who solves reliability first.
tweet
RT @godofprompt: 🚨 BREAKING: OpenAI just hired the creator of OpenClaw, the most viral AI agent of 2026.
Peter Steinberger will lead the next generation of personal agents. Sam Altman called him “a genius.”
I see it differently.
OpenClaw went viral. It also got flagged by Gartner as a cybersecurity risk, went rogue on users’ iMessage accounts, and had its own maintainer warn people not to use it if they can’t run a command line.
Manus hit $100M ARR without a proprietary model. OpenClaw spawned an entire agent ecosystem overnight.
Neither solved the hard problem: making autonomous agents reliable at scale.
Error recovery at step 14. State management across sessions.
Permission boundaries that stop your agent from emailing your entire contact list.
The wrapper layer is not the moat. The reliability engineering layer is.
Models are commoditizing. The value just migrated to the teams who make unreliable systems actually work.
There are maybe three teams on Earth who’ve proven they can do that.
OpenAI didn’t acquire an agent platform. They acquired a vision they still need to engineer into something trustworthy.
The real AI agent war isn’t who builds the flashiest demo. It’s who solves reliability first.
Peter Steinberger is joining OpenAI to drive the next generation of personal agents. He is a genius with a lot of amazing ideas about the future of very smart agents interacting with each other to do very useful things for people. We expect this will quickly become core to our product offerings.
OpenClaw will live in a foundation as an open source project that OpenAI will continue to support. The future is going to be extremely multi-agent and it's important to us to support open source as part of that. tweet
X (formerly Twitter)
Sam Altman (@sama) on X
Peter Steinberger is joining OpenAI to drive the next generation of personal agents. He is a genius with a lot of amazing ideas about the future of very smart agents interacting with each other to do very useful things for people. We expect this will quickly…
The Transcript
$IAC CEO: Google Search referrals down 50% over the last 2 years
"Looking at the core sessions, we're down 13% year-over-year on the quarter. The biggest contributor to that is a 50% drop in Google Search referrals over the last two years. This quarter, we also saw a little softness in non-search traffic sources, mainly driven by declines in Google Discover, which is their version of Apple News, which had been a contributor to non-search growth earlier in the year."
tweet
$IAC CEO: Google Search referrals down 50% over the last 2 years
"Looking at the core sessions, we're down 13% year-over-year on the quarter. The biggest contributor to that is a 50% drop in Google Search referrals over the last two years. This quarter, we also saw a little softness in non-search traffic sources, mainly driven by declines in Google Discover, which is their version of Apple News, which had been a contributor to non-search growth earlier in the year."
tweet
Dimitry Nakhla | Babylon Capital®
RT @DimitryNakhla: 20 Quality Stocks Down ≥20% From Highs — Now in Their Own Bear Markets 🐻
1. $ISRG -21%
2. $TDG -21%
3. $MCO -21%
4. $NDAQ -21%
5. $AMZN -22%
6. $MELI -24%
7. $MSFT -26%
8. $SPGI -27%
9. $BKNG -29%
10. $UBER -30%
11. $BSX -31%
12. $SNPS -32%
13. $RACE -34%
14. $PLTR -37%
15. $BMI -39%
16. $NFLX -43%
17. $FICO -44%
18. $APP -47%
19. $NOW -54%
20. $CSU -55%
tweet
RT @DimitryNakhla: 20 Quality Stocks Down ≥20% From Highs — Now in Their Own Bear Markets 🐻
1. $ISRG -21%
2. $TDG -21%
3. $MCO -21%
4. $NDAQ -21%
5. $AMZN -22%
6. $MELI -24%
7. $MSFT -26%
8. $SPGI -27%
9. $BKNG -29%
10. $UBER -30%
11. $BSX -31%
12. $SNPS -32%
13. $RACE -34%
14. $PLTR -37%
15. $BMI -39%
16. $NFLX -43%
17. $FICO -44%
18. $APP -47%
19. $NOW -54%
20. $CSU -55%
tweet
Offshore
Photo
God of Prompt
RT @rryssf_: Tencent researchers found a way to get reinforcement learning performance without updating a single parameter
it costs $18. the RL methods it outperforms cost $10,000+
the method is called Training-Free GRPO, and the core idea is more interesting than the cost savings https://t.co/krYfosYQLj
tweet
RT @rryssf_: Tencent researchers found a way to get reinforcement learning performance without updating a single parameter
it costs $18. the RL methods it outperforms cost $10,000+
the method is called Training-Free GRPO, and the core idea is more interesting than the cost savings https://t.co/krYfosYQLj
tweet