Kapisch's avatar
Kapisch
kapisch@nostr.fan
npub1l0wk...mwpp
Stay humble and keep stacking sats⚡️ Creator of SkyLocation App Available now on App Store
Kapisch's avatar
BitbyBit 5 months ago
Pizza Napoletana is the only way 🤌 image
Kapisch's avatar
BitbyBit 5 months ago
Adopt a Verification Loop when using AI LLMs sound confident so be systematic: – Ask for sources & uncertainty levels – Run a quick web check or second-model check Trust but verify. Accuracy is a habit, not a feature.
Kapisch's avatar
BitbyBit 5 months ago
The fastest way to learn AI in 2025: 1️⃣ ChatGPT → Learn fundamentals & practice prompting 2️⃣ Perplexity → Research + stay updated w/ citations 3️⃣ Claude → Master long docs, reasoning & writing 4️⃣ GitHub Copilot → Hands-on coding partner (learn by doing) 5️⃣ HuggingFace → Train & deploy real open-source models 6️⃣ Kaggle → Free datasets + practical ML challenges 7️⃣ YouTube / Fast.ai → Bite-sized lessons from pros AI isn’t learned from theory… it’s learned from building. 🚀
Kapisch's avatar
BitbyBit 5 months ago
Me in Public keeping my opinions to myself.
Kapisch's avatar
BitbyBit 5 months ago
Peeping Monkey. A cool idea for a login screen, where the eyes follows you type and then covers its eyes when you type the password. #Vibecoding
Kapisch's avatar
BitbyBit 5 months ago
I just discovered that prompting in Chinese saves huge tokens vs English. Same idea, fewer tokens → more context + lower cost. But keep in mind that doing this might lose context when translating and there maybe less trained data on your topic too. Still a cool fact though. image
Kapisch's avatar
BitbyBit 5 months ago
When the waitress who said they'll be right back with your sauce, is walking around the restaurant living her best life.If this isn’t the most beautiful UI experience, I don’t know what is! #ios26
Kapisch's avatar
BitbyBit 5 months ago
Use AI daily? Try keeping a Prompt Log: – Write down every good prompt you use – Include: Goal • Model • Example output • Notes – Tag by use-case: writing, coding, analysis, marketing, etc. In 30 days, you’ll have a personal prompt library.
Kapisch's avatar
BitbyBit 5 months ago
Want smarter outputs from AI models? Inject reasoning with just 4 words: “Let’s think step by step.” This tiny phrase hacks the model into mimicking more deliberate logic. #AI #vibecoding
Kapisch's avatar
BitbyBit 5 months ago
Morning view from top of the hill. Past days have been really hot, so making the most of the cold mornings!
Kapisch's avatar
BitbyBit 5 months ago
Good target to have on social media. image
Kapisch's avatar
BitbyBit 5 months ago
Thinking about getting into AI ? My first basic advice is create a dedicated new email ID for all AI model/newsletter sign-ups. This keeps all AI-related stuff in one clean inbox so nothing gets lost.
Kapisch's avatar
BitbyBit 5 months ago
image While using AI think in tokens and not words. Your prompt is not read letter by letter. It's chopped into tokens. “unbelievable” = [“un”, “believ”, “able”] That’s why spelling, formatting, and syntax matter more than you think.
Kapisch's avatar
BitbyBit 5 months ago
There are over 500,000 AI models you can use right now but most people only know ChatGPT. Here are 5 worth trying today: 1️⃣ Claude 🔗 claude.ai 2️⃣ Gemini 🔗 gemini.google.com 3️⃣ Mistral 🔗 mistral.ai 4️⃣ LLaMA 3 🔗 huggingface.co/meta-llama 5️⃣ Perplexity 🔗 perplexity.ai which one is your go to AI model for daily drive?
Kapisch's avatar
BitbyBit 5 months ago
Nothing is as important as you think it is as you are thinking about it.
Kapisch's avatar
BitbyBit 6 months ago
If there is a cup I hate on this planet, its this one! I mean wtf? image