Uploaded July 2026 | Updated November 2026, 1 day ago
In this video, Riaz Virani, Enterprise Engineer at Roboflow, demonstrates the latest additions to Vision Events, including Operator Feedback, the built-in Event Dashboard, custom dashboards, and natural-language queries through the Roboflow Model Context Protocol (MCP) Server.
When you deploy a computer vision model across many sites, the operators on the floor often know things the model does not. Historically, capturing that judgment has meant custom side systems, standing calls with distributed teams, or spreadsheets that do not scale. Operator Feedback closes that gap by capturing feedback directly in Vision Events, so the events your team flags as incorrect become the data that improves the next model version.
Riaz also walks through the built-in Event Dashboard for a quick view of what's happening across your deployments, shows how to build a fully custom dashboard on top of the Vision Events API, and demonstrates how the Roboflow MCP Server lets you query your Vision Events data in plain English.
= Chapters =
0:00 Introduction: Better Models and Insights in Vision Events
1:14 The Gap Between the Model and the Floor
3:41 Operator Feedback: Battery Cell HMI Demo
5:41 Filter for Incorrect Events and Add Them to Training
7:12 Closing the Loop with Operator Feedback
8:26 The Built-In Event Dashboard
11:14 Custom Dashboards with Sign In with Roboflow
12:31 Natural-Language Queries via the Roboflow MCP Server
13:34 Wrap Up: Getting Data Into Vision Events
= Resources =
Vision Events launch blog: blog.roboflow.com/vision-events
Operator Feedback documentation: docs.roboflow.com/deploy/vision-events/operator-feedback
Query Events documentation: docs.roboflow.com/deploy/vision-events/query-events
In this video, Riaz Virani, Enterprise Engineer at Roboflow, demonstrates the latest additions to Vision Events, including Operator Feedback, the built-in Event Dashboard, custom dashboards, and natural-language queries through the Roboflow Model Context Protocol (MCP) Server.
When you deploy a computer vision model across many sites, the operators on the floor often know things the model does not. Historically, capturing that judgment has meant custom side systems, standing calls with distributed teams, or spreadsheets that do not scale. Operator Feedback closes that gap by capturing feedback directly in Vision Events, so the events your team flags as incorrect become the data that improves the next model version.
Riaz also walks through the built-in Event Dashboard for a quick view of what's happening across your deployments, shows how to build a fully custom dashboard on top of the Vision Events API, and demonstrates how the Roboflow MCP Server lets you query your Vision Events data in plain English.
= Chapters =
0:00 Introduction: Better Models and Insights in Vision Events
1:14 The Gap Between the Model and the Floor
3:41 Operator Feedback: Battery Cell HMI Demo
5:41 Filter for Incorrect Events and Add Them to Training
7:12 Closing the Loop with Operator Feedback
8:26 The Built-In Event Dashboard
11:14 Custom Dashboards with Sign In with Roboflow
12:31 Natural-Language Queries via the Roboflow MCP Server
13:34 Wrap Up: Getting Data Into Vision Events
= Resources =
Vision Events launch blog: blog.roboflow.com/vision-events
Operator Feedback documentation: docs.roboflow.com/deploy/vision-events/operator-feedback
Query Events documentation: docs.roboflow.com/deploy/vision-events/query-events










