A Workspace Picker, Agent Bug Reports, and a Clearer Checkout

Three changes shipped today that you will run into quickly: one when you open an org, one the next time an agent hits an error, and one when you pick a plan. A later release made listings, search and shares faster. Here is what changed, plus the smaller fixes that went out with them.

Opening an Org Starts With a Workspace Picker

Entering an org now takes you to a Choose a workspace picker.

• Recent workspaces come first, each with its member and agent count, followed by the full list.
• Your workspaces are separated from the ones available to join, so you can tell at a glance where you already belong and where you could.
• Browse opens the same picker, and the create-workspace flow has a Back button, so you can leave it without creating anything.

If your org runs a lot of workspaces, with people in some and agents in others, this is the screen that makes the difference.

Agents Can File Bug Reports

When an agent runs into something broken, it can now say so. The MCP server has a new report-bug action, and the API has a matching endpoint, so an agent or a person can file a platform bug report from the place the failure happened.

The report is only useful if we can find what went wrong, so two things support it:

• Failed calls now come back with a request ID. An agent can quote it in the report, and so can you.
• A report can be linked to the matching server error, so it can arrive with the error behind it rather than only a description of symptoms.

A Clearer Checkout

Checkout was rebuilt so you can see what you are paying for before you pay it.

• Two columns. Plan details sit on one side and an order summary on the other, including what you owe today.
• An interrupted payment is not a dead end. If a payment gets cut off and you sign back in, you return to the checkout you were in instead of starting over.

Plan details live on the pricing page.

Faster Listings, Search and Shares

The later release cut down how much work the server does for each request. You will notice it most in big folders and busy workspaces.

• Storage, search and member lists make far fewer database queries.
• Folder listings fetch previews and virus-scan status in one batch, which keeps large folders quick to open.
• Search and metadata lookups load files in larger batches, so big result sets need fewer round trips.
• Shares respond faster. Recording folder-share activity now happens in the background instead of while you wait.

Smaller Changes

• One name for Intelligence. The label that used to read High Intelligence is gone, and the app uses the same single name throughout.
• The browse tray and the folder tree work together. Files in the tray’s list view can now be dragged straight onto a folder in the tree.
• OAuth consent is clearer. The “All” access option is hidden when it cannot be granted, and the errors explain what went wrong.
• Closing an org reports failure honestly. If cancelling the subscription behind it fails, closing the org fails too, instead of claiming success.
• Expired file shares are cleaned up again. The cleanup job had been failing quietly on every run. It now does its job.
• Large storage listings no longer crash if a cache connection drops partway through.
• The app no longer loads files from outside CDNs. Icons, the Domine typeface and the PDF viewer now come from Fastio, so a strict content security policy or network allowlist no longer needs exceptions for Iconify, Google Fonts or unpkg.
• Event summaries are complete. They had been returning partial results.
• Folder-share activity records who triggered it, so the audit trail shows which user acted.

The workspace picker is the change most people will see first. The bug-report path is the one we expect to matter more over time: an agent that can tell us what broke, with a request ID attached, is an agent we can help.