What's New

This production update includes customer-facing improvements from the latest release.

What changed

• Reliability and performance improvements are included.
• The landing page flow, trust information, and legal links are easier to scan.

This production update includes customer-facing improvements from the latest release.

What changed

• Report creation, editing, and export workflows are more reliable.
• The landing page flow, trust information, and legal links are easier to scan.

This production update includes customer-facing improvements from the latest release.

What changed

• The landing page flow, trust information, and legal links are easier to scan.

This production update includes customer-facing improvements from the latest release.

What changed

• Refine Aidana workflow UX and staging pipeline.
• Workspace file handling is clearer and more dependable.
• Add Azure Foundry provider support.
• The landing page flow, trust information, and legal links are easier to scan.
• Business settings and team configuration are more dependable.
• Report creation, editing, and export workflows are more reliable.

This production update includes customer-facing improvements from the latest release.

What changed

• The landing page flow, trust information, and legal links are easier to scan.
• What's New now respects language selection and avoids duplicate navigation.

We spent the last two days closing several bugs and tidying things under the hood. That makes Aidana more predictable, and gets us ready for marketing.

What you will notice

• Deleting agents and skills now works as expected. Previously, a deleted agent could show up again on the next load or leave "ghost files" in your workspace folder. Now the file is moved to a .trash/ folder, so it is recoverable if you change your mind, and the live folder stays clean.
• Default skills keep their formatting when you open and save them from the workspace tree. A quiet bug could previously cut off parts of the Formatting template in popular templates like Project Status Report. That is fixed.
• Per-agent model selection for Business users now actually works. If you have configured a specific model or key for the writer or critic, it is used in the pipeline. Previously the setting was read, but ignored.

A little under the hood

• We removed server-side data storage that is no longer needed now that your vault is the canonical source. Less surface area, more in line with the "everything in your own vault" principle.
• We ran the first real stress test against the test environment: 22,495 requests over 5 minutes, with 0 errors. The stack has comfortable margin for 500 concurrent users.

Tell us if anything looks odd. We are here.