Choose a model for your cleaning scripts

Fredit Flow connects to a cloud AI provider through a standard API endpoint, to a model running locally on your PC, or to the built-in model. Privacy differs for each. Your choice decides where the prompt text goes, which matters for files with personal details.

Three ways to connect a model

A cloud AI provider

A cloud model connects through a standard API endpoint and suits larger scripts and harder rules. The provider may charge separately. The prompt text, including any file text the agent reads, goes to that provider, so keep files with personal details out of it unless you have the right to send them.

A model running locally on your PC

A local model keeps every prompt on your machine. Nothing leaves the PC. That makes it the choice for CSV files with names, e-mails or ages. It runs on your own hardware, and a small local model may write weaker scripts, so the diff review matters even more.

The built-in model

The built-in model is the option for trying Fredit Flow without connecting a cloud provider or running a model yourself. Try sample files first. For personal data, choose a model running locally on your PC, where nothing leaves the machine.

Check the connection with Test

The Test button checks an endpoint before you start. It shows the response time and the context window, which tells you how much text the model can take in one request, for example a sample of a long file plus your script. Test every new model.

Pick a model by the size of the job

Use a bigger model for refactors and complex logic, such as a script that merges several tables, and a smaller one for small edits like trimming a column. The run_on_model tool sends one hard step to a bigger model while the rest stays on your usual one. Decide before sending data.

Quality varies. Read the diff before you accept. Ask first pauses before actions that change something, and the real exit code and output show whether a script ran, not whether its result is right.