Tricentis Tosca Agentic AI: Autonomous Software Testing

Tricentis Tosca Agentic AI

Autonomous Automation for the Enterprise

The Innovation Bottleneck

Software delivery speed is increasing exponentially.

Traditional manual testing cannot keep pace.

Defining Agentic AI in Testing

Autonomy

AI that plans, executes, and adapts without constant human prompts.

Context Aware

Understands complex business logic, not just code syntax.

Generative Test Creation

From natural language to fully functional automation.

From Natural Language to Automation

End-to-End Workflow of Tosca Agentic AI

1. Input: Natural Language

Users provide high-level intent via simple text prompts.\n\n<b>Example:</b> "Login as 'Admin' and verify the dashboard loads."\n\nNo technical scripting or code required—just business logic description.

2. AI Processing Engine

A sophisticated multi-modal engine orchestrates the creation:\n\n• <b>Context Agent:</b> Uses LLMs to parse semantic meaning, user intent, and workflow logic from the prompt.\n• <b>Vision Agent:</b> Analytically scans the application UI to identify buttons, fields, and structures without relying on brittle IDs.\n• <b>Strategy Agent:</b> Determines the optimal execution path, bridging the gap between human intent and technical steering parameters.

3. Output: Tosca Test Case

The AI automatically builds a structured, model-based test case.\n\n• Includes all steps, steering params, and validations.\n• Ready to run, resilient, and self-healing.

Democratize Quality

Empowers business users and manual testers to create robust automation without coding skills.

Accelerate Productivity

Reduces test creation time by up to 90% by instantly converting prompts into test cases.

Enhance Resilience

Generates self-healing models that adapt to UI changes, minimizing maintenance efforts.

Leveraging External LLMs for Script Creation

Workflow: Generating Tosca Test Cases from CSV Skeletons

1. Generate CSV Skeleton

Engineers prompt external LLMs (e.g., GPT-4, Gemini) to generate high-level test steps in a structured CSV format. The LLM acts as the creative content generator based on requirements.

Step,Action,Value\nLog In,Input,User123\nVerify,Check,Welcome

2. Utility Conversion

A simplified utility (e.g., Python script) parses the CSV. It maps 'Actions' to standard Tosca TBox modules and 'Values' to TestStepValues using a predefined mapping logic.

def convert_to_tbox(row):\n if row['Action'] == 'Input':\n return TBox_SendKeys(...)

3. Tosca Script Import

The utility generates a Tricentis subset (.tce) or uses the API to ingest the tests directly into Tosca Commander. The result is a fully executable, modular test case.

<Subset>\n <TestCase name='Login'>\n <TestStep.../>

Vision AI & Self-Healing

Standard scripts break when UI elements move. Tosca's Vision AI sees the screen like a human, identifying buttons and fields regardless of underlying code changes.

Risk-Based Optimization

Tosca prioritizes tests based on business risk, achieving high coverage with fewer tests.

Model-Based Automation

Decouples technical layer from business logic.

Reusable modules reduce redundancy.

Updates once, reflects everywhere.

Seamless CI/CD Integration

Agentic AI orchestration fits directly into your DevOps pipeline, triggering tests upon every commit.

Industry Benchmarks: Efficiency

Start testing 85% faster.

Future-Ready Quality Assurance

Accelerate digital transformation with Tricentis Tosca Agentic AI.