We should be able to run the AI part of the program without needing the API to simplify the development. This should make measuring the outcomes easier.
There should be a test which makes sure the local dev pipeline still works - test should use mocked llm clients to not spend tokens
We should be able to run the AI part of the program without needing the API to simplify the development. This should make measuring the outcomes easier.
There should be a test which makes sure the local dev pipeline still works - test should use mocked llm clients to not spend tokens