QA agents that write test cases
n8n AI agents that generate test cases from Trello cards, flag duplicate bugs as they are reported, and answer questions about the system.
- Role
- QA Engineer and AI & Automation Engineer, Tidal Information Systems.
- Period
- 2024 – 2026
- Status
- Used by a QA team
- Stack
- n8n, LLM agents, Trello API
- ~40%QA effort saved (estimate)
- 50%fewer production bugs (with a traceability process)
Architecture
System steps in order
- Test cases
- Trello card (Input)
- AI agent: generate test cases (AI)
- Test cases for the team (Output)
- Duplicates
- New bug report (Input)
- LLM compares it with existing bugs (AI)
- Possible duplicate? (Alert)
- Notify reporter (Output)
- Knowledge base
- Team question (Input)
- Retrieve system docs (Data)
- AI answer (AI)
The problem
Testers wrote every test case by hand from Trello cards, and the same bugs were often reported more than once by different people.
What I built
Three n8n AI agents: one generates test cases from a Trello card, one compares each new bug report with existing bugs using an LLM and flags likely duplicates, and one answers team questions about the system as a knowledge base.
Results
An estimated 40% of the testing team's time saved; together with a traceability matrix and mentoring 3 testers, production bugs fell by 50%.