How are completed flows scored by AI?

What is it for?

Every completed flow is scored 0–100 by AI against the criteria you define. Each score comes with its reasoning and how reliable the underlying data is. For example, use it to pre-assess applicants in an application flow.

Results appear only on the flow's “Analytics” tab; the person who filled in the flow never sees them.

Turning on evaluation

1. Open the “Evaluation” tab in the flow editor.
2. Turn on “Turn on evaluation for this flow”. While it's off, the criteria are kept but completed flows aren't scored.
3. Click “Add criterion”. Write the criterion's name (for example Financial health) and “What to look at and how to score it”: say which answer or document field to look at, and what earns a high or a low score.
4. Add more criteria if needed, then click “Save”.

If evaluation is on but there are no criteria, flows aren't scored. The flow can't be saved while a criterion has no name or instructions.

Reference documents and links

You can add optional “References” to each criterion, such as your organization's scoring guide:

• “Upload document (PDF, DOCX, TXT, MD)”: up to 10 MB.
• “Reference link”: enter the address and click “Fetch”.

The reference's text goes to the model with the instructions; the file itself isn't stored. Delete a criterion with “Remove criterion”; the live flow doesn't change until you save.

Reading the results

Evaluation runs in the background once the flow is completed. Results are listed in the “Evaluations” view on the “Analytics” tab: the average score, the score distribution and, for each conversation, the score per criterion with its reasoning and confidence (“Confirmed by data”, “Partial / self-reported”, “No data”). The overview also shows the “Avg. fit score”.

Criteria that couldn't be scored are listed separately. If evaluation is off, Analytics shows “Evaluation is off” with a “Set up evaluation” link back to this tab.