This release adds Gemini support, makes workflow debugging substantially easier, and ships a complete platform redesign. We're also sharing an early look at our mobile app.
Native Gemini Support
Gemini is now available across the platform. Access Google's models alongside your existing providers - useful when you need multimodal capabilities or want to benchmark performance across different model families.
Available models: Gemini 2.5 Pro, 2.5 Flash, and 2.5 Flash Lite
To enable Gemini, go to your organization's model settings and add your API credentials.
Memories in Prompts
Prompts can now reference memories as attachments. This makes it straightforward to build prompts that pull from organizational knowledge without manually copying context. Define your memories once, attach them where needed, and they'll stay current as the underlying data changes.
Improved Workflow Debugging
Variable Store - Workflows now include a variable inspector that shows your global state during execution. Instead of adding debug nodes or guessing where data transforms went wrong, open the variable store and see exactly what's there.
Cleaner Run Mode - Workflow execution is now easier to follow. We've tightened up the visual feedback and state management so you can see what's running and where things are in the process.
Files in Playgrounds
The Playground now handles file and memory attachments. Test prompts with actual documents before committing them to production workflows. This closes a gap where you'd previously need to build a workflow just to see how a prompt performed with file inputs.
Visual Refresh + Mobile App (soon)
We've refreshed the entire interface. Navigation is cleaner, information hierarchy is clearer, and the visual design is more consistent.
Next up - We're building a mobile client for the platform. The focus is on accessing assistants and monitoring workflow runs when you're away from your desk.
Reply to this email if you want to join the beta.