Talk
Cheap to Write, Expensive to Run: AI Testing and Token Costs
AI agents can write and run tests in minutes, but every run has a token bill, and it grows quietly with each sprint. This session looks at where tokens actually go in AI-driven test automation, why a test that is cheap to author can become the most expensive one to run, and whether there is a way to optimise the usage.
Key takeaways
- How AI shifts testing costs from creation to execution.
- When to use AI agents, scripts or a hybrid.
- Where AI agents waste tokens, and how to cut costs.
- Why cheaper testing isn't always better testing.