How this problem bothered me... About half a year ago, I was hired as a prompt engineer and didn't have a service to run prompt tests at my company. We were a startup not looking for a huge, expensive solution.

I needed this testing tool to allow me to: SHOW MY RECEIPTS. Evidence of why I choose to use a certain LLM provider (price, results) over another for a task. MAKE INFORMED DECISIONS. For personal projects, an easy way to see which language model would be the best choice for a given task.

How I built glideprompt? I started building this solo (using just node and react) and realized there was even more opportunity to simplify the prompt-engineering process: 1. Stop writing system prompts from scratch, it writes me a boilerplate one -for the task- to start with. Oftentimes, staring at a blank slate wastes so much time in any area of life, including coding or prompt engineering. 2. Saving the prompt tests for later on firestore, so I can use it again when new open source models come out, etc.

I figured other people would have the same problem so I tried to make the interface as simple as simply freaking possible. Moving forward, I'm looking to see if anyone wants to use this as an API connection, or would be more interested in running larger tests, etc.

I really want to make this a collaborative customer/founder thing. Comment or email and I guarantee I will respond, and take your advice/feedback seriously.

Please let me know what you think.