Prompt optimization
Improve prompts by testing variations against your criteria and keeping the ones that score better.
What is Prompt optimization?
Prompt optimization proposes changes to a prompt, tests each version on your data, and keeps the versions that score better on your criteria. It turns prompt writing into a measured process. Spro's prompt optimization is planned and not yet available.
Prompts are usually improved by hand, one guess at a time. Optimization automates the loop: propose a change, test it on a set of examples, keep what works and repeat.
This page describes planned behaviour. The optimization method and cost of each run will be published with the feature.
Key features
What this feature does for your requests.
- Proposes prompt variations
- Tests each variation on your examples
- Keeps versions that score better
- Works with your own criteria
How it works
4 steps, from setup to the first request.
- 01
Start from a prompt
Use the prompt you want to improve.
- 02
Provide examples and criteria
Define what a good answer looks like.
- 03
Run the optimization
Variations are generated and scored.
- 04
Review the best version
Accept or reject the proposed prompt.
Is it the right fit?
Specific situations where this feature helps, and where it does not.
Good fit if
- A prompt is important and has a clear measure of success
- You have examples and an evaluation set to test variations
- A manual rewrite has stopped improving results
Not a good fit if
- The evaluation set is small or unrepresentative, so gains may not generalise
- The prompt rarely changes and manual tuning is enough
- Success cannot be measured
Prompt optimization FAQs
Common questions about Prompt optimization.
Route every AI request through SPRO.
One OpenAI-compatible API for 350+ models, with intelligent routing, 29 built-in guardrails, observability, and automatic failover.
