This month we're shipping improvements across connector capabilities, credential obfuscation, and agent usability. February's updates expand where you can write data, give admins tighter control over how API secrets are stored and referenced, and reduce prompt setup overhead in Vision Agent.
​

We believe powerful tools should advance alongside the analysts who depend on them. February improvements deliver smoother processes, stronger trust in your data, and the clarity required to make every insight count.
​

Vision Agent now includes a template dropdown directly in the node configuration. Select a document type, like bank statements, invoices, or purchase orders, and get a structured extraction prompt populated in the node, ready to run or adjust.

This reduces the prompt setup effort significantly. Instead of writing field definitions, output structure, and multi-page handling from scratch, you start from a proven baseline and edit what you need. Savant handles structural guardrails behind the scenes.

File connector metadata already includes timestamps, paths, and filenames, but to actually open the source file, you had to reconstruct the URL manually or search the source system.

Savant now adds a fileUrl field to metadata for every record processed through file-based connectors. A direct link to the file in Box, OneDrive, SharePoint, or Google Drive. No breaking changes, no added latency.

API tool credentials lived directly in the tool config, visible to anyone who opened the node. Multiple flows meant the same key copy-pasted across nodes with no central way to rotate or revoke.

Admins can now store credentials in Savant's secure Vault. Analysts reference them via ${vault.SECRET_NAME}. Values are obfuscated in the UI after saving and resolve only at runtime. Only Admins can create, update, or delete secrets.

The Fabric connector supports Warehouse read-write via SQL, but Lakehouse Delta tables have been read-only since the SQL analytics endpoint doesn't support writes.

Savant now adds Spark Mode to the Fabric connector, executing Spark jobs directly in your workspace. Create new Delta tables, append, or overwrite, directly from your pipeline destination. Same OAuth flow, job status in the run UI, and existing SQL Mode connections are completely unaffected.

These updates expand your connector reach, strengthen credential obfuscation, and simplify agent setup. We're continuing to ship improvements every week, all designed to help you move faster, work smarter, and deliver your best work with less effort.

Have feedback, ideas, or requests? We’d love to hear from you.