Browse Comments — Raw (as collected)

RAG is the layer most people underestimate. The brain is only as useful as what it can actually see.

Game-changer!

I see this confusion often in leadership conversations, Luís. People talk about AI as one thing, when it’s really multiple systems working together.

MCP determines whether intelligence remains trapped inside isolated tools or becomes interoperable across systems through a unified connection layer. Without standardization at this level, integration complexity grows into long term technical debt, Luís.

Nice, but seven days of doing real tasks beats seven months of watching someone else use the tool. The fastest Claude users I know never even finished a tutorial.

day 5 is where claude stops being a chatbot and people never get there because they quit after day 2

Joe Prasen, we're exactly the same :) i like Cowork and Obsidian combo, its especially helpful in editing the files

The body analogy works better than most AI explainers I've seen. The teams I talk to have a brain, half a library, no nervous system, and they're confused why nothing moves coherently.

Strong focus on personalization through writing samples and banned-word lists. That step alone often decides whether outputs feel generic or usable.

Digiaxo, appreciate the share :)

I hope they found a reliable Ai that doesn’t hallucinate at all. Sounds like a mess

day 3 is exactly where most people skip and then wonder why Claude sounds generic... teaching it your voice first changes everything.. been doing this with claude.md files at work and the difference is night and day. Ruben Hassid

This is basically the modern “onboarding arc” for AI tools, not learning features, but building an actual system around them.

Well said, Prem. Enterprise AI depends on how well these layers are connected.

Ruben --- I would like to know how to start with claude poc for financial systems- first claude.ai; then claude cowork and then claude code

Ruben, keeping it simple wins. People spend hours learning tools they’ll never fully use. A basic repeatable setup gets you further. Building your own takes a bit up front, pays off every week.

Fabiola Singh, those who start day 1, even badly, will learn more in 20 minutes than 3 hours of YouTube you can't learn better than actually using the tool

Most people are still watching tutorials about the tool instead of actually building something with it. the gap between those two groups is only getting wider.

Chandan Ray, thanks for supporting :) anything else you would add?

AI only becomes powerful when it’s trained on your actual workflow, not tutorials. Ruben Hassid