This week we’ve focused on making Spark more reliable and responsive when building your apps.

Enhanced reliability

We’ve resolved several issues that were interrupting your workflow:

  • Fixed errors that prevented agent iterations from saving properly
  • Resolved initialization failures that blocked new agent iterations from starting
  • Improved handling of multiple tool calls to speed up generations

These fixes mean fewer interruptions and a smoother experience when iterating on your apps.

Prepopulated seed data

When you create a new app, Spark now automatically populates your data store with relevant seed data. You’ll see a realistic preview of how your app looks and functions with actual content, helping you visualize the end result and iterate faster.

Screenshot of Spark app preview showing populated seed data in a table format

Additional improvements

  • Users now see clear notifications when Spark experiences high demand, with automatic failover to alternative models so you can keep building even when our primary model is at capacity.
  • Uploaded video assets now properly display in the app preview.
  • Published app links correctly open in new tabs.
  • Updated repository creation flow with confirmation dialog for easier access to created repositories.

Try these improvements today

Visit github.com/spark to build your next app with enhanced reliability!