Skip to content

Video Intelligence Agent

Transforming Test Creation from Video to Automation-Ready Assets Video Intelligence Agent logo

What is Video Intelligence Agent?

Video Intelligence Agent is a goal-oriented and autonomous decision-making QA agent on Google Agentic AI (Vertex AI/Gemini) that converts recorded video walkthroughs into standardized production-ready BDD test cases to transform test creation from video to automation-ready assets.

It is an AI-powered agent built on the A2A Protocol v1.0. It accepts application demo videos and automatically generates production-ready Behavior-Driven Development (BDD) test cases in Gherkin syntax. Under the hood it leverages Google's Gemini multimodal AI model to watch your video, understand application workflows, and produce structured test output covering:

  • Happy path scenarios — the primary success flows
  • Negative scenarios — what happens when things go wrong
  • Boundary condition tests — edge cases and limits
  • Data-driven tests with Scenario Outlines and Examples tables

Built for Google Cloud Run and designed to serve tens of thousands of concurrent users.

Key Features

Feature Description
Video Analysis Accepts MP4, WebM, and QuickTime video formats
Structured Output Returns organized Gherkin .feature files grouped by business domain
A2A Protocol Fully compliant — interoperable with any A2A-compatible client or platform
Real-time Streaming Live status updates and incremental delivery of each feature file as it is generated
Cloud-Native Deployed on Google Cloud Run — auto-scales for high concurrency
Comprehensive Tagging Auto-tags each scenario with priority, type, and nature (@P1, @smoke, @positive, etc.)

How It Works at a Glance

sequenceDiagram
    participant User
    participant VideoIntelligenceAgent as Video Intelligence Agent
    participant Gemini as Gemini AI

    User->>TestMorph: Send video (+ optional text context)
    TestMorph-->>User: Acknowledged — Analyzing video…
    TestMorph->>Gemini: Forward video for multimodal analysis
    Gemini-->>TestMorph: Structured BDD output
    TestMorph-->>User: Feature file 1 (e.g. login.feature)
    TestMorph-->>User: Feature file 2 (e.g. checkout.feature)
    TestMorph-->>User: Summary (total features, scenarios, flows)
    TestMorph-->>User: Completed

Technology Overview

Component Technology
Language Python
AI Model Google Gemini 2.5 Pro
Protocol A2A v1.0
Deployment Google Cloud Run