If you’re 42 then your probably were 16 or 17 years old in 2000. You probably weren’t in tune with platforms trying to sell themselves as decentralized.

100%. After a month or two of seriously diving into agent assisted coding, I can honestly say it’s the elusive 10x’er to productivity management has always dreamed about.

Everywhere I can. Let’s say I need to build a feature, take for instance a common task of integrating with Stripe. I’ll start by gathering a few links to api documentation and use the Claude chat to say something along the lines of “research integrating with Stripe for invoicing referencing this links. Create a detailed design spec for the features” Then I’ll take that spec, give it a once over and save it to my code workspace. I’ll execute Claude Code and prompt something like “Think about the design spec for integrating stripe and create a detailed implementation plan to integrate Stripe in this application staying true to the projects conventions”. I’ll give that a once over and maybe even ask it to validate the plan against the original spec. Once I’m satisfied with the spec I say “execute the plan”. I then move to a different project while it’s running and repeat back and forth. I’ve been very happy with the quality of code given enough context and guidelines.

Claude Code is a great agent for coding. The Claude Sonnet models are also superior to pair with your agent.

Claude Code