
Scopic Devs Prompt Smarter with Claude
How our devs saved time and code using AI, step by step
4 followers
How our devs saved time and code using AI, step by step
4 followers
Vibe coding with Claude 3.5 sounded like a dream... until it broke. This article unpacks how we improved results using structured prompting. A dev-first look into where LLMs help, where they fail, and what actually works in real workflows.




We recently ran a dev experiment at Scopic using Claude 3.5 Sonnet to speed up our internal tool development.
We hit common pitfalls (“vibe coding” burnout, loops, broken logic), but found simple ways to prompt better and build faster.
This write-up is from our dev Mladen and shares what actually worked for us, with prompt examples and workflow notes.
Would love to hear how others are using LLMs in coding. Drop your tips or horror stories below.
Here is the Medium article → https://scopicsoftware.medium.com/we-built-an-internal-tool-90-faster-using-ai-heres-what-we-learned-0558f05d64cd
Here’s the whitepaper that inspired it → https://nextcloud.scopicsoftware...