Promting Advice
A developer is seeking advice on improving their AI prompting techniques, as their current inefficient workflow sometimes exhausts their usage limits. They have built a .js system for hardware control with a UI and commercial clients. Initially coding themselves, they now rely on AI for complete code generation, feature implementation, updates, and rollouts. They also use AI, specifically Astra 6, for PCB design and production, often having new AI models review the work of older ones, which contributes to heavy usage due to suboptimal prompting.
Time & source
Times shown in UTC
Display time zone: UTC
Local time zone unavailable; showing UTC.
PublishedOffset at this time: UTC+0Sep 12, 2026, 07:19 UTC
IngestedOffset at this time: UTC+0Sep 12, 2026, 12:01 UTC
- Published
- Sep 12, 2026, 07:19
- Ingested
- Sep 12, 2026, 12:01
- Source type
- Dev community
- Tier
- Community
- Source status
- Healthy
Tier is a per-source editorial setting, not a per-item score.
Hello!
I got told that my prompting might suck and that’s the reason I burn trough my limit sometimes.
I’d like to get some advice on how to prompt properly.
I build a .js system that uses Sensor Data to steer various hardware and offers a pleasing UI to do so.
I already got some commercial clients. As I started to work with AI I still coded shit my self, but it got to a point where the AI output was way ahead of my skills. Right now AI basically writes the complete code, I just got the Ideas of what to code.
Since I started with like reviewing my own Skripts and promting for that I got used to an now inefficient AI Workflow that becomes more and more insufficient.
Right now I use AI to implement new features, update features and keep them running. I also do roll outs with AI. I usually let a new Model Review the work of the old one (usually very heavy usage sure cause of bad prompting).
Astra 6 does/ assist me in PCB Design and Produktion.
Glad for advice, thanks! :)