In this release, we’ve got major updates for the Agent Hub and model connection, new modifiable Contributor permissions, and more!

Agent Hub updates

Agents in Orchestrate

Models that you connect through the Agent Hub can now be used as Agent nodes in your pipelines in Orchestrate. You can link them to events or actions so that you can integrate models into your automation.

Additionally, whenever an Agent is set off, you can view its details in Monitoring as you would with any Action, and you can filter by its data too.

New providers

Our Agent Hub now supports models from a list of new providers. The parameters and specifications for these models must be configured within a JSON schema during connection. The providers we’ve added are:

• Fireworks AI
• Vertex AI
• Databricks
• AWS Bedrock

Additional output components

You may now select additional component types when selecting the output component for a model: Number, Slider, Rating, and Voting.

This applies to Multimodal projects, for both the Run Model button component in the form builder, and the Run Model feature in Explore.

New parameters in Explore

When using the Run Model feature in Explore, you may now define additional parameters depending on the type of model selected. These parameters are Reasoning effort and Verbosity.

Contributor permissions

Team Owners and Team Admins can now provide a list of management permissions to contributors, allowing them to carry additional responsibilities within their team. These permissions include:

• Manage Contributors’ permissions - This provides the rest of the permissions to a contributor while allowing them to do the same for other contributors as well.
• Invite Contributors to team
• Remove Contributors from team
• View Contributors’ custom field values
• Edit Contributors’ custom field values

New component functions

• We’ve added two new functions for the Web component:
  • getItemStatus()
  • getUserRole()

What do you think of the updates? We'd love to hear your feedback.