@moritzkremb Fully agree! Been using it for lots of PoCs and internal process improvements. Very useful. Also scales suprisingly well for most use cases.
πΏπ¦ Free tax calculator with medical aid tax credits!
β Latest SARS rates
β Medical aid tax credits
β Age-related rebates
β No signup needed
Perfect for tax season prep (opens 20 July!) Try it now: https://t.co/qTtXy4BRkV
#TaxCalculator#SARS#MedicalAidTaxCredit
O yeah! I missed the boat on @AnthropicAI 's notification about Google docs π integration to Claude. This will be a big step forward for many orgs.
π
Windsurf vs Cursor
Haven't tried Windsurft, but it looks like a good alternative at a considerably lower pricepoint.
Anyone with experience of switching who can weigh in on what you liked or didn't?
https://t.co/YrjInusvay
@bindureddy I think it's a marketing/naming problem. If they're right about it's capability (I haven't had a chance to experiment much) then it should have released it as Cinquain 3.5 and made a minor incremental increase to Haiku and nobody would have batted an eyelid at the price jump.
@imrat Definitely prefer Perplexity here even if it means waiting for a few more seconds. The chart is a nice touch, but also the answer to the question (the actual current share price) is MUCH more visible.
Wondering π€ after my weekend #AI coding fun with my daughter... We're constantly asking LLMs to explain things. How many "learning by explaining" opportunities are going lost in the process? Is this a hidden cost of AI? #FutureOfWork#Learning#Collaboration
Had a blast this weekend coding a swimming race game with my 8-year-old using @cursor_ai composer! It's a bit glitchy, but seeing her excitement as our dolphin π¬ dodged obstacles and sharks was great. #AI#coding#familyfun#learntogether
@AnthropicAI's computer use π₯οΈ feature on Claude looks really interesting. Where RPA π€ meets LLM βοΈ. Potentially a lot of efficiency to be gained here and many, many use cases that spring to mind in my day-to-day. Definitely going to give it a test run π.
π€π€― OpenAI's Whisper is crushing it in speech-to-text! π₯ Google and AWS are lagging behind, and Anthropic? Nowhere to be seen. π€
Can they close the gap to @OpenAI if they can't nail this? #speechtotext#AI#knowledgemanagement#OpenAI#Whisper
@awscloud have been quietly plugging away at improving Bedrock and it's now my goto for any LLM workloads. The @AnthropicAI partnership seems to be thriving and they have so much more (incl. Bedrock studio that's a hidden gem!)
My pet hate π‘ for @OpenAI remains the lack of any API support for cost and usage reporting.
The worst part is, they have an api for this, they just don't want people to use it π
Why?
Interesting thread with @cursor_ai team's insights. Definitely agree with the take that 3.5 sonnet is top model for coding rn.
I find o1 useful occasionally, but on first iteration it will often introduce unnecessary breaking changes.
I just watched the first hour of the Lex Fridman podcast with the Cursor team.
I've put together 10 of my favorite moments from it and snipped the sections of the podcast below.
Let me know if you want the 2nd hour too.
TLDR; one of the best deep dives into coding with AI. It's not just for Cursor fans. Anyone interested in coding with AI should watch.
Whether you use Claude, o1, Aider, GPT Engineer, Copypasta, Amazon CodeWhisperer, or Replit Agent, there's something here for you.
Did I miss anything? What was your favorite bit?