Skip to main content
Hello from ARGOS Omni. Here is what shipped in August.

Grounds for every verdict

New Feature An analysis result now gives you more than values — it tells you why each value came out that way. The Report tab shows the reasoning field by field.
Report tab

The Report tab — each field carries the reason behind its value.

Reasons are attached per field, not per step. When one step filled in several fields, each field gets its own reason. In the API, every leaf field arrives in an envelope with three keys — value, reasoning, and sourceStep — see reading analysis results.
Analyses that completed before this feature existed have no reasoning, and it is not generated retroactively. “No reasoning provided” on screen does not mean there were no grounds — it means the reason could not be supplied.

Why a field came back empty

New Feature A field defined in your output schema that received no value is now marked not extracted on the Report tab, together with why it is empty. This tells you whether to request a better document, fix the policy, or look at an engine problem. It also explains a low Output Completeness in the Analysis Score.

Edit reference notes

New Feature The reference notes attached to each action are now editable. Use Edit Reference Notes on the workflow edit screen, one entry per line.
Edit Reference Notes

Edit Reference Notes — one line is one entry.

Changes apply without regenerating the playbook, so there is no 2–3 minute wait. Up to 20 entries per action, 500 characters each.
When a verdict does not match your intent, check the reference notes before rewriting the policy. The cause is usually not the policy sentence itself but that sentence being carried into the action’s reference notes incorrectly.

Descriptions and metadata on folders and items

New Feature Folders and items now take a description and metadata of their own. Attach an identifier from your system and it comes back unchanged as metadata in the API response.

More AI models to choose from

Feature Improvement Four more AI models are available at step 3 of workflow creation — AI model and engines.