The "happy path" is a lie—especially when there’s a Large Language Model involved.
Most AI implementations look great in a controlled demo, but they crumble the moment they hit
unpredictable user inputs or high-concurrency production loads. If you aren't actively trying to
break your AI-generated code, your users eventually will.
If you don’t know your system’s breaking point, the universe will assign one to you.
In this workshop, we will build integration and performance tests for a small distributed system.
We will explore integration tests:
1. How AI can make mocking parts of the system easy
If you can define the testing scenario, it will be built
2. Human in the loop and verification are very important for tests. 3. How can we utilize agents
to minimize the human-in-the-loop?

We will explore for performance tests:
1. What are the types of performance tests?
2. What are the best in-class performance testing tools?
3. How to build and maintain performance tests

Workshop Details

Participants will build integration and performance tests for a small distributed system.
Participants should have access to AI-assisted IDEs and plugins such as Claude Code, Cursor,
Copilot, Antigravity. Free tiers are acceptable as well, however best-in-class tools do make a
difference.
The workshop’s focus will be on building an AI agent that will assist in producing both integration
and performance tests.
Finally, the workshop will explore how these tests can be integrated into the SDLC