I eye AI | Making the most of what AI has to offer | Always looking for the next big thing | Follow to keep an ๐๏ธ on the latest Tools, Tutorials & Prompts.
Anthropic: "modest but tangible improvement"
41 days since last Opus.
Same price. 88.6% SWE-bench.
3ร cheaper fast mode.
Honest about what it doesn't know.
The pace isn't slowing.
It's compounding.
Where do you think the ceiling is?
Claude Opus 4.8 is really bad at handling money.
I tested this today with 6 live naked option shorts sitting in my Quantplay account. It calculate the Greeks, and understand your risk perfectly.
And then refused to close a single one.
I ran the exact same setup in Hermes. It did a dry run first, showed me every BUY order it was about to place, asked for confirmation, and then closed all 6 positions cleanly.
Same account. Same positions. Same instructions. Completely different outcome.
When you're building with AI agents, the question you should be asking is not which model scores highest. It's which one will actually take action when connected to something real.
A lot of these models are incredibly good at analysis and incredibly hesitant to do anything with it. That's not a useful agent, that's just a smart chatbot with extra steps.
Screenshots below. ๐
๐ Building something big!
Not just a URL shortener โ a one-stop solution for:
โ URL shortening
โ Link tracking & analytics
โ Click heatmaps
...and much more!
Need a killer .store domain name for this.
Drop your best ideas below ๐
#URLShortener#LinkTracking#StartupIdeas
Anthropic: "modest but tangible improvement"
41 days since last Opus.
Same price. 88.6% SWE-bench.
3ร cheaper fast mode.
Honest about what it doesn't know.
The pace isn't slowing.
It's compounding.
Where do you think the ceiling is?
๐ Day 1 of building my first SaaS!
The idea: A URL shortener โ but way more than that.
Here's what I'm building:
๐ Smart URL shortening
๐ Click tracking & analytics
๐จ Custom QR code generation
Building in public. Sharing every win, every bug, every lesson.
Let's go. ๐งต๐
๐ Day 2 of System Design โ Leaderboard System
Following @Striver_79's System Design Sheet and today hit the
Leaderboard problem.
Seems simple right? Just sort users by score.
But at 50 Million DAU โ nothing is simple.
Here's what I learned today:
๐ญ,๐ฏ๐ญ๐ณ ๐ฐ๐ผ๐ป๐๐ฟ๐ถ๐ฏ๐๐๐ถ๐ผ๐ป๐ ๐ถ๐ป ๐ฎ๐ฌ๐ฎ๐ฒ โ and the year isn't even half over.
Consistency beats intensity, every time. Green squares don't lie.
What does your streak look like? ๐
#Coding#GitHub#BuildInPublic#SoftwareEngineering
๐ Day 2 of System Design โ Leaderboard System
Following @Striver_79's System Design Sheet and today hit the
Leaderboard problem.
Seems simple right? Just sort users by score.
But at 50 Million DAU โ nothing is simple.
Here's what I learned today:
@striver_79 Started with a basic 2-box diagram. Ended with write path, read path, realtime notifications, WebSocket server, message queues and cron workers.
That's the difference between knowing system design and actually understanding it.
Day 3 coming tomorrow ๐