BTC TestAgent

Product Preview

The agent drives. You stay in review.

An AI-driven virtual tester that automates system-level HIL testing end to end - and uses our formal, symbolic methods to make the AI's results trustworthy.

Book a Demo

Agent WorkflowAwaiting Review

Requirements Extracted

Requirements Formalized

Test Cases Generated

Symbolically Validated

Executed on HIL

RequirementsSystem modelSILHIL

System-level test generationPowered by BTC QCore

BTC TestStack

Code-level · Unit & integration · Engineer drives

BTC TestAgent

System-level (HIL) · Agent drives · Trusted

Why This Exists

System-level testing is still mostly manual.

System-level and HIL (Hardware-in-the-Loop) testing today is largely manual, knowledge-intensive and slow. Engineers translate requirements into tests by hand, set up execution, and judge results case by case.

Generative AI promises to accelerate this - but on its own it isn't trustworthy enough for safety-relevant verification: results can be incomplete or plausible-looking yet wrong.

BTC TestAgent closes this gap. Generative AI brings speed and automation; proven formal methods bring confidence and quality - moving system-level testing toward a 24/7 automated activity.

How It Works

Full system-level test workflow.

You provide the specification documents. The agent takes it from there, step by step, with you in control wherever it's needed.

Requirement Extraction

Reads specification documents and extracts the relevant requirements automatically.

Requirement Formalization

Turns informal requirements into formal, machine-verifiable requirements using our Universal Pattern technology.

Test Case Generation

Derives coverage items and generates test cases in a multi-step process, including automated completeness and correctness checks.

Test Case Validation Loop

Generated test cases are validated symbolically before execution. Findings feed back into generation, so incomplete or incorrect test cases are corrected automatically.

Test Execution

The validated test cases are executed against the system under test in the HIL / simulation environment.

Result Verification

Execution is evaluated automatically against the formalized requirements, delivering a clear verdict with requirement coverage and full traceability.

Test Execution Feedback Loop

Execution results and the resulting verdict feed back into the workflow, so the agent learns from the system's real behavior and improves the next run.

Built-In Capabilities

Automatic where it can be. Manual where it should be.

Human-in-the-Loop by Design

Steps run automatically when the agent's confidence is high. When confidence is low, you're prompted for the missing information and can guide the step manually.

Manual Unlocking of Steps

Inject additional instructions or skip a step by providing your own inputs, so the workflow never gets stuck.

Two Feedback Loops

The validation loop and the execution loop let the agent learn from both verification findings and real execution verdicts - within a run and across runs.

Bring Your Own Data

Feed in your own process- and system-specific documents to experience the agent in your own setting.

What Makes It Distinctive

Not AI instead of rigor. AI made rigorous.

Core differentiator

Trusted AI - AI Plus Formal Methods

BTC TestAgent enhances generative AI with our symbolic methods to make the AI's output verifiable and trustworthy.

Universal Pattern Technology

Our formalization technology expresses requirements as precise, verifiable patterns - the basis for both symbolic test-case validation and automatic verdict evaluation.

Confidence-Driven Automation

The agent acts autonomously where it is sure and asks where it is not, combining automation speed with engineering judgment.

Built on BTC QCore

Formalization, symbolic validation and verdict evaluation run on BTC Embedded Systems' formal-verification technology, proven in safety-critical projects for years.

Integrated with the dSPACE Test Automation SDK

Test creation and execution build on dSPACE's abstract API for HIL test creation and execution, aligned with modern SDV developer tooling.

What You Should Know

Built to fit into how you already work.

Local Install: Runs on your local machine - a low installation hurdle that fits your existing environment.
Cloud-Ready: Prepared for the cloud - designed to scale into your cloud and HIL environment over time.
Your Own LLM: Works with the LLM access your company already has (currently OpenAI-class models).
Existing Know-How: Reuses your existing HIL testing knowledge, rather than forcing a new methodology.
Your Own Data: Add your process- and system-specific documents to try it in your real context.

Try It First

Comes with a ready example.

A Brake Light Controller example with a virtual simulation lets you experience the full workflow immediately, without needing your own data first.

Book a Demo