For engineering, test and QA teams
AI in development & test: with Claude Code, OpenAI Codex and GitHub Copilot.
From engineers who build test systems themselves. Your team learns to build its own tools with AI, and stays in control.
$ claude
› Derive test cases from REQ-112 and write pytest tests for them.
✓ 14 test cases derived, 3 edge cases flagged
✓ tests/test_req112.py created
✓ 14 passed in 0.41s
› Summarise this as a test report.
✓ reports/REQ-112_test_report.mdWhere AI really helps in engineering.
Reviewing requirements
Find gaps and contradictions early.
Deriving test cases
From requirements, with traceable links.
Test bench software
Test scripts and evaluations in Python.
Modernising legacy code
Understand and renew it, for example LabVIEW to Python.
Test reports
From bullet points and measurement data, in your template.
Your own tools
Small internal tools in days instead of weeks.
The tools at a glance.
As of October 2026, vendor-independent.
- Claude Code used dailyAI agent in the terminal: reads your repository, writes code and runs tests.
- OpenAI Codex used dailyThe OpenAI counterpart, also as a desktop app and in the cloud.
- GitHub CopilotAI right in the editor, good for a gentle start.
- Open modelsOn your own servers, when code must not leave the building.
How we work.
- 1
Consultation
30 minutes by video, on a real problem from your daily work.
- 2
Kickstart
3 online sessions: setup, first tool, ground rules.
- 3
In the team
Workshop and hands-on project in your code base.
Also for regulated environments.
We come from medical device and automotive development ourselves.
A human checks
Every AI contribution is reviewed before it counts.
Traceable
Everything runs through your version control.
Data by traffic light
Sensitive code stays in-house.
Your QM rules
For example under IEC 62304 or ISO 26262.
What we build with it ourselves.
This website was built with Claude Code. In our engineering business Liquidbeam Experts we use the same tools:
- End-of-line test systems and test automation
- Calibration software for medical devices in Python
- Firmware for STM32 and ESP32
- Migrating grown LabVIEW test benches
Quick answers.
May we give our source code to Claude Code or Codex?
In the business and enterprise plans, the vendors state that your data is not used for training by default. For particularly sensitive code, there are open models on your own servers. The decision lies with your IT security and data protection teams.
Will this replace our developers?
No. It takes away routine work. Knowledge, architecture and responsibility stay with your team.
Does this also work with embedded code or LabVIEW?
Very well with C, C++ and Python. With LabVIEW, AI mainly helps with understanding, documenting and migrating to Python.
Free consultation
Show us a real problem.
30 minutes by video with screen sharing, in English or German. We show you what AI makes of it.