OpsFlow Dashboard is a static prototype of a lightweight AI-assisted operations dashboard.
It is designed for an operations associate (named "Meng") to track onboarding workflows,
follow-ups, and process improvements. The UI is clean and structured, rendered as a single
scrollable page inside the TERMINAL_VIEWER iframe.
Observed Behavior
The top of the page displays a header bar labeled "OpsFlow Dashboard" with a brief description and the logged-in user ("Meng, Operations Associate").
An Onboarding Metrics section shows three summary stat cards: 5 Avg Launch Days, 12 Open Tasks, and 2 Blocked Tasks. These are static hardcoded values.
A Tasks section lists two items with owner, due date, priority, and status badges. "Provision software access" is marked in_progress; "Confirm device shipment" is marked blocked in red.
An AI Assistant panel is partially visible at the bottom of the viewport, suggesting it continues below the fold.
All content appears to be static — no interactivity was observed (no clickable tasks, no editable fields, no live AI queries).
Things That Did Not Work as Expected
The app is entirely static. Despite being described as "AI-assisted," no AI functionality is wired up. The AI Assistant section is a placeholder with no input or response capability.
Task items have no interactivity — clicking a task does nothing. There is no way to update status, reassign ownership, or mark tasks complete.
The metric cards (Open Tasks, Blocked Tasks) do not reflect or link to the actual task list shown below — the numbers appear hardcoded and disconnected from the data.
Due dates listed (2026-03-01, 2026-03-02) are already in the past relative to the current date, with no visual overdue indicator.
No navigation, filtering, or sorting is available for the task list, which would be essential at any realistic scale.
Improvement Prompt
You are building OpsFlow, a lightweight AI-assisted operations dashboard. Improve the current static prototype with the following changes:
1. Make the task list interactive. Each task should have a dropdown to update its status (not_started, in_progress, blocked, done). Changing a status should update the Onboarding Metrics counters in real time.
2. Add an overdue indicator. If a task's due date is before today, highlight it in red and add an "OVERDUE" badge.
3. Wire up the AI Assistant panel. It should accept a plain-text question about the current tasks (e.g. "Which tasks are blocked and who owns them?") and return a short, contextual answer based on the actual task data on the page. Use a language model API call for this.
4. Add a simple "Add Task" form with fields for task name, owner, due date, and priority. New tasks should appear in the list and update the metrics immediately.
5. Keep the visual design minimal and professional. No unnecessary decoration — just clean typography, clear status badges with distinct colors (blue for in_progress, red for blocked, green for done, gray for not_started), and a responsive single-column layout.