[Payment due @huult] Render server-driven AZ status on expense reports - #95959
Conversation
…iven-AZ-status-on-expense-reports # Conflicts: # src/pages/inbox/report/ReportActionsListTailIndicator.tsx # tests/unit/ReportActionsListTailIndicatorTest.tsx
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 056f875ba5
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
Reviewer Checklist
Screenshots/VideosAndroid: HybridAppScreen.Recording.2026-07-21.at.15.34.37.movAndroid: mWeb ChromeScreen.Recording.2026-07-21.at.15.39.27.moviOS: HybridAppScreen.Recording.2026-07-21.at.15.43.42.moviOS: mWeb SafariScreen.Recording.2026-07-21.at.15.52.08.movMacOS: Chrome / SafariScreen.Recording.2026-07-21.at.15.19.32.mov |
joekaufmanexpensify
left a comment
There was a problem hiding this comment.
Good for product 👍
|
@inimaga Could you help set up your local env so I can access it to test this PR? |
|
@huult You don't need the additional setup. The backend PR is already deployed so you can test directly against staging. In the instructions for manual test above, you can go straight to test steps. |
Codecov Report❌ Looks like you've decreased code coverage for some files. Please write tests to increase, or at least maintain, the existing level of code coverage. See our documentation here for how to interpret this table.
|
|
🚧 inimaga has triggered a test Expensify/App build. You can view the workflow run here. |
This comment has been minimized.
This comment has been minimized.
Screen.Recording.2026-07-21.at.12.46.46.movYeah, I can see it. I'll do some more testing and update the checklist. |
Screen.Recording.2026-07-21.at.12.58.44.movReproduction Steps
Current behavior: Expected behavior: @inimaga Could you take a look at this issue? |
Untitled.2.mov@inimaga I occasionally encounter this issue. After sending a request, Concierge remains in the "Thinking..." state and never returns a response. Retrying the same request usually succeeds: Screen.Recording.2026-07-21.at.13.24.43.mov |
Screen.Recording.2026-07-21.at.13.33.13.mov
Current behaviorAfter Concierge finishes processing, only the McDonalds expense is updated to "Lunch". The second request is not processed, and Concierge does not post the expected acknowledgement message for it. Expected behaviorAfter the first request completes, Concierge should also process the second request (or clearly indicate that it cannot). An acknowledgement message should be posted for each successfully completed request. @inimaga could you check this issue? |
|
@inimaga Could you take a look at my comment when you have a chance? |
|
@huult Sorry for the delay here. Regarding the flagged issues; For issue 1: Send an initial message to change the description, and a subsequent message referencing the old changed description.
The fix for this belongs in the backend. Thanks for flagging. I will create a separate issue to track this. The App PR only renders the status indicator so I want the potential blockers here to only be focused on this. For issue 2: Occasional stuck thinking state. Looking at the videos you submitted, its not actually stuck in a thinking state (as there is no thinking indicator). But rather, it stopped thinking, but there wasn't an acknowledgment to say what it decided to do. In this particular instance, a simple acknowledgement would have sufficed. To be also fixed in the backend. Will track with a separate issue. For issue 3: Follow-up request while Concierge is thinking is not processed. This isn't actually limited to this flow and is an issue with Concierge handling multiple requests even in the main flow. When you start a concierge 1-on-1 dm and ask it a question, then while its thinking, ask another question, it'll only answer / action your first question. But overall, I agree this makes for a bad product experience so will make another separate issue to fix across the product (not just this flow). Overall we shouldn't block on the above 3 issue as they are beyond the scope of this PR. Nevertheless, I will be addressing them separately in follow-ups. |
|
@cristipaval Please copy/paste the Reviewer Checklist from here into a new comment on this PR and complete it. If you have the K2 extension, you can simply click: [this button] |
|
🎯 @huult, thanks for reviewing and testing this PR! 🎉 A payment issue will be created for your review once this PR is deployed to production. If payment is not needed (e.g., regression PR review fix etc), react with 👎 to this comment to prevent the payment issue from being created. |
|
Trackings for backend follow-up issues created here: |
|
Thank you |
|
✋ This PR was not deployed to staging yet because QA is ongoing. It will be automatically deployed to staging after the next production release. |
|
🚧 cristipaval has triggered a test Expensify/App build. You can view the workflow run here. |
|
🧪🧪 Use the links below to test this adhoc build on Android, iOS, and Web. Happy testing! 🧪🧪
|
|
🚀 Deployed to staging by https://github.com/cristipaval in version: 9.4.44-0 🚀
|
Help site review — no docs changes requiredI reviewed the changes in this PR against Why: This PR is internal UI plumbing — it lets non-chat (expense) reports mount the AgentZero status gate when the server writes a processing-indicator NVP, so parent multi-expense reports can render the server-driven Concierge "thinking"/status bubble. That indicator is transient, in-product visual feedback; it introduces no new user-facing feature, setting, tab, label, button, or workflow that the help site documents. The actual user-facing capability this supports — editing/updating an expense by chatting with Concierge on a report — is already documented and accurate in If a future PR adds a documented behavior change (e.g., a new Concierge capability, a user-visible setting, or copy the help site references), flag it and I'll draft the docs update then. @inimaga, please confirm you agree no help site changes are needed here. If you'd still like a docs PR created (or believe an article should call out this behavior), reply with the details and I'll open one. |
|
🤖 Payment issue created: #97137 |
|
🚀 Deployed to production by https://github.com/marcaaron in version: 9.4.44-6 🚀
Bundle Size Analysis (Sentry): |
Explanation of Change
Held on:
This PR updates the AgentZero status provider so non-chat reports can mount the status gate when the server has written an AgentZero processing indicator NVP, allowing parent expense reports to show server-driven thinking/status bubbles. Expense reports without server status still skip optimistic kickoff, and unit coverage now verifies both the server-driven render path and the quiet default path.
Fixed Issues
$ https://github.com/Expensify/Expensify/issues/651801
PROPOSAL: N/A
Tests
Manual tests
Pre-requisite (Skip to Test Steps if testing in/with Staging/Prod):
Test Steps:
“Lunch with James” to the McDonalds expense on this report.
Expected result:
Concierge should emit/show a thinking indicator while the request is in progress. At the end of which, the description update succeeds and the parent multi-expense report receives a visible Concierge acknowledgement.
Issue.651801.-.Manual.Test.mp4
Offline tests
QA Steps
Same as manual test steps above
PR Author Checklist
### Fixed Issuessection aboveTestssectionOffline stepssectionQA stepssectionAvatar, I verified the components usingAvatarare working as expected)StyleUtils.getBackgroundAndBorderStyle(theme.componentBG))npm run compress-svg)Avataris modified, I verified thatAvataris working as expected in all cases)Designlabel and/or tagged@Expensify/designso the design team can review the changes.mainbranch was merged into this PR after a review, I tested again and verified the outcome was still expected according to theTeststeps.Screenshots/Videos
Android: Native
Android: mWeb Chrome
iOS: Native
iOS: mWeb Safari
MacOS: Chrome / Safari