Building AI for leverage | Thinking out loud
Joined January 2018
- Tweets8.9K
- Following173
- Followers385
- Likes1.3K
The higher the stakes, the more efficiently we tend to solve problems.
When the stakes are low, we brute-force our way through. When they’re high, we look for the optimal solution from the start.
Which means most people/things are operating far from their efficiency ceiling simply because they can afford to.
Best use of AI is to pictorially represent greatest historical events, we have enough real events to produce GOT level visuals, or even better #̢̛ͥ̀̽̽ͨ̌̓͆̇̀̽ͦͨͫͯ̏̎̇̈̚͜҉̕͢͏͡҉̸̶̛͢͡҉̵̨̭̳̘̹̫͔͖̜̠͖̫̲̖̱̤̻̖͉͢͞͡ͅͅ͏̸̷̢͑̊̀̐ͣͭͭ̉͂ͨ̽͒̊̎͛ͪ̊̎ͩͧ̀͝͠͠͏̸̵̵̷̵
Got a call from AWS with an offer to migrate to it from Azure, with 3x credits for 3 months of what I spend on Azure. Is it worth it?
I just downgraded to premium from premium plus, and I'm seeing a burst of ads instantly
Can't even imagine what a basic plan would be like
Fugu is basically Japan’s perplexity in an IDE and has SDK support
Introducing Sakana Fugu: A full multi-agent orchestration system accessible via a single model API.
Our ‘Fugu Ultra’ model matches the performance of Fable and Mythos, delivering frontier capability without the risk of export controls.
Try it: sakana.ai/fugu 🐡
Interviewer :
Why does zoom have offices?
Him :
They work from office so everyone else can work from home
We got fable 6 before fable 5
Introducing Sakana Fugu: A full multi-agent orchestration system accessible via a single model API.
Our ‘Fugu Ultra’ model matches the performance of Fable and Mythos, delivering frontier capability without the risk of export controls.
Try it: sakana.ai/fugu 🐡
When you know you don’t know, you pay more attention, and then you finally know what you didn’t know.
Founders, there’s a lesson for you.
OBSESSION
Snapchat just unveiled AR glasses called Specs at $2,195 a pair
Their stock $SNAP dropped 10% immediately
The glasses run fully standalone so no phone is needed by using two Qualcomm chips hand tracking AI assistant
Spiegel told CNBC people are ready to think about computing differently after 20 years of staring at a phone screen
Snap tried smart glasses in 2016 and they flopped
Now they are back with a new version trying to go up against Meta and Google
Either Spiegel is a genius who is 10 years ahead of everyone
Or has he just spent a decade making the same mistake twice? 😬👓
Asked fable 5 with ultracode effort to refactor my codebase
It made 295 tool calls, wrote millions lines, created 27 files, wrote thousands unit tests and modularized everything with surgical precision.
Zero of it compiles
10/10.
Art doesn’t have to work but it’s beautiful.
Mayank Khanna retweeted
Hiring these days is an investment for the future. It has no immediate value, cz Claude Code is better than most engineers, but companies understand the pattern: IQ is going to drop significantly.