Manual prompt engineering can take weeks of trial and error and still fail to arrive at optimal prompts for a given application. Even worse, once you've tuned your prompts for a specific model, migrating or upgrading to another model requires you to start the prompt engineering process over from scratch.

Prompt adaptation solves this problem by automatically generating optimized prompts for each of the models you're considering working with, improving performance without protracted manual effort.

When to use prompt adaptation

Prompt adaptation is ideal when you need to:

• Improve accuracy - Drive accuracy improvements of 5-50%+ over your workflows
• Migrate between model providers - Switch from OpenAI to Anthropic, Google, or other providers
• Evaluate multiple models - Test which model works best for your use case with optimized prompts for each
• Reduce vendor lock-in - Maintain flexibility to switch providers without rebuilding your application
• Improve cost, latency, and security - Optimize prompts for cheaper, faster, and open source models

By automating the prompt engineering process, you can outperform manual prompt engineering in minutes instead of weeks.