up2parts – AI-Powered Manufacturing Software
Building a production-grade E2E automation framework for an AI co-pilot where output accuracy directly impacts production decisions.
Service
QA & Automation, Test Infrastructure
QA & Automation, Test Infrastructure
Clients
up2parts GmbH
up2parts GmbH
Date
2025
2025
Software
autoCAM, Playwright, Cucumber, Azure DevOps
autoCAM, Playwright, Cucumber, Azure DevOps
Location
Europe
Europe
The Challenge
up2parts is an AI co-pilot that automates work preparation for machining companies — generating CAM programs, work plans, and cost calculations. When AI output feeds directly into production decisions, errors create scrap, rework, and delays. The challenge was designing and delivering a production-grade E2E automation framework that could validate correctness across the platform and scale with the product.
Industry
Manufacturing, CAM / CNC, Work Preparation
Manufacturing, CAM / CNC, Work Preparation
Scope
E2E Test Automation, Visual Regression, CI/CD Integration
E2E Test Automation, Visual Regression, CI/CD Integration
Role
End-to-end Quality Engineering Consultancy
End-to-end Quality Engineering Consultancy
Validation role
End-to-end Quality Engineering Consultancy
End-to-end Quality Engineering Consultancy
What was being validated
- AI-generated CAM programs and work plan outputs
- 3D Part Viewer visual correctness across machining sides
- Two-step Auth0 authentication including organisation picker handling
- Session management under repeated test execution
- ERP and calculation integration workflows
Validation focused on output correctness, visual consistency, and long-term automation reliability.
The Approach
A fully fledged E2E framework was designed and built using Playwright, Cucumber, and TypeScript with Page Object Model architecture. Functional tests covered login flows and negative paths. Visual regression tests captured the 3D Part Viewer canvas per machining side, compared against pixel-level baselines with diff output. Session clearing, credential management, and CI/CD integration were built in from the start, with XRAY-compatible reporting providing full scenario-to-ticket traceability.
Quality Outcomes
- Production-grade functional and visual test coverage delivered end to end
- Stable, repeatable execution with automatic session management
- XRAY-compatible reporting with scenario-to-ticket traceability
- Azure DevOps pipeline configured for CI-triggered execution
- Framework designed to scale across multiple product modules and environments
Why This Matters
Avento’s role
In AI-driven manufacturing software, quality is not just about the platform working; it is about trusting the output enough to act on it. When AI suggestions directly influence production decisions, automation must be rigorous, repeatable, and built in from the start.
Building AI-powered software where output quality matters?
Let’s make it reliable end to end.
