sagarpatil

6 ms·

31.
by sagarpatil 1y ago
I always use Claude Code to debug issues, there’s no point in trying to do this yourself when AI can fix it in minutes (easy to verify if you write tests first) o3 with new search can do things in 5 mins that will take me at least 30 mins i
32.
by sagarpatil 1y ago
Use free tiers?
33.
by sagarpatil 1y ago
Yeah, you can’t do sh*t in an hour. I spend a good 6-8 hours every day using Claude Code, and I actually spend an hour every day trying new AI tools, it’s a constant process. Here’s what my today’s task looks like: 1. Test TRAE/Refact
34.
by sagarpatil 1y ago
What really baffles me is the claims from: Anthropic: 80% of the code is generated by AI OpenAI: 70-80% Google/Microsoft: 30%
35.
by sagarpatil 1y ago
Let need explain. Windows user for 28 years. I decided to try macOS and got hooked (no more WSL). Moved from Android to iPhone because of tight integration with macOS. 1. Universal copy (copy from/to Mac) 2. Automatically fill OTP fro
36.
by sagarpatil 1y ago
The worst part is I’m so deep in Apple ecosystem that it’s impossible to move to Android.
37.
by sagarpatil 1y ago
Meanwhile Apple: Liquid Glass
38.
by sagarpatil 1y ago
Have they ever decreased the price in the past? I don’t remember.
39.
by sagarpatil 1y ago
Before: 50 messages per week Now: 100 messages per week
40.
by sagarpatil 1y ago
That has been my conclusion too but how do you explain the long ass prompt by AI labs: https://docs.anthropic.com/en/release-notes/system-prompts#m...
41.
by sagarpatil 1y ago
Older than United States.
42.
by sagarpatil 1y ago
So your sample size is 1 task and 1 LLM? I would recommend trying o3, opus 4 (API) with web search enabled.
43.
by sagarpatil 1y ago
Just bought it. Any advice on how to use the book to master the terminal? One page a day or try to finish everything in a week or two?
44.
by sagarpatil 1y ago
The worst part is, these scams flourish during bull markets, like now, when alts start shooting up 10-20% everyday. Stay safe.
45.
by sagarpatil 1y ago
I avoid using my own API keys, especially for Sonnet 4 or Opus, because LLMs can rack up unexpected costs. Instead, I use Augment Code’s remote agents and Google’s Jules, which charge per message rather than by usage. This setup is ideal fo
46.
by sagarpatil 1y ago
Cool story.
47.
by sagarpatil 1y ago
Sorry. I was wrong.
48.
by sagarpatil 1y ago
Sonnet 3.7 > Sonnet 4? Interesting.
49.
by sagarpatil 1y ago
Did you not see the live stream? They took a feature request for excalidraw (table support) and Claude 4 worked on it for 90 minutes and the PR was working as expected. I’m not sure if they were using sonnet or opus.
50.
by sagarpatil 1y ago
You can’t in AI studio, it’s available on loveable.
51.
by sagarpatil 1y ago
It allows you to use your own api ke.
52.
by sagarpatil 1y ago
4.1 in api already provides 1 million tokens. Anthropic’s enterprise version does too. I’m not sure if this is a software or a hardware (computer) problem.
53.
by sagarpatil 1y ago
They are referring to SWE bench lite. Just want to make sure you are too.
54.
by sagarpatil 1y ago
Why are you obsessed with Pelicans? What’s your story?
55.
by sagarpatil 1y ago
Why not use Claude Max Plan? Starts at $100.
56.
by sagarpatil 1y ago
I signed up on the waitlist when it was announced, got my invite today.
57.
by sagarpatil 1y ago
I don’t understand the hype in the original post. OpenAI launched function calls two years ago and it was always possible to create a simple coding agent.
58.
by sagarpatil 1y ago
It generates a PR. You decide if you want to merge, clone it locally, test, edit and then merge the code.
59.
by sagarpatil 1y ago
I am working with a 3rd party API (Exa.ai) and I hacked together a python script. I ran a remote agent to do these tasks simultaneously (augment.new, I’m not affiliated, I have early access) Agent 1: write tests, make sure all the tests pa
60.
by sagarpatil 1y ago
Re:simulation Deebo does this for debugging: https://github.com/snagasuri/deebo-prototype

More ›