Quality assurance · Production
Assembled saves hundreds of engineering hours by using LLMs to generate tests
The problem
Writing comprehensive tests is often skipped due to time constraints or complexity, forcing engineering teams to trade off development speed against code quality.
Workflow diagram · grounded in source
1
Craft prompt with code context
trigger
“you should craft a precise prompt that guides the model to produce the desired output”
2
LLM generates test suite
ai_action
“we fed this code into ChatGPT o1-preview and, in just 48 seconds, it generated a comprehensive test suite that was ready to use straight out of the box”
3
Engineer reviews and refines
human_review
“you might need to review and refine the generated tests. You should check for compilation issues, add any potential edge cases the LLM missed, and adjust the style to match your codebase conventions”
4
Tests merged into codebase
output
“We insist that all Assembled engineers read and run any LLM-generated tests before merging into production”
Reported outcome
Assembled engineers collectively saved hundreds of hours, individual engineers saved weeks of time, and tasks that previously took hours are now completed in 5–10 minutes.
Reported metrics
Time per test task5–10 minutes
Individual engineer time savedsaved weeks of time
Collective engineering hours savedhundreds of hours
Test suite generation time (example)48 seconds
Reported stack
o1-previewClaude 3.5 SonnetChatGPTCopilotCursor
Source
https://www.assembled.com/blog/how-we-saved-hundreds-of-engineering-hours-by-writing-tests-with-llms
Read source ↗Frequently asked questions
What did this team achieve with this AI workflow?
Assembled engineers collectively saved hundreds of hours, individual engineers saved weeks of time, and tasks that previously took hours are now completed in 5–10 minutes.
What tools did this team use?
o1-preview, Claude 3.5 Sonnet, ChatGPT, Copilot, Cursor.
What results were reported?
Time per test task: 5–10 minutes; Individual engineer time saved: saved weeks of time; Collective engineering hours saved: hundreds of hours; Test suite generation time (example): 48 seconds (source-reported, not independently verified).
How is this quality assurance AI workflow structured?
Craft prompt with code context → LLM generates test suite → Engineer reviews and refines → Tests merged into codebase.