Turn AI curiosity into engineering confidence.
"Everyone's talking about AI agents, but how do I actually test one?"
"I keep hearing about MCP servers and multi-agent systems. Where do I even start?"
"My team is shipping GenAI features and I'm expected to have an opinion on quality. I don't yet."
If any of that sounds familiar, you're not alone. GenAI agents are showing up in more and more products, and the testing community is being asked to make sense of them fast. So Jaswanth Manigundan built a course to help.
How to test GenAI Agents (by building one) is now live in the MoTaverse, created by Jaswanth Manigundan (Jas), Engineering Manager at Culture Amp.
Rather than talking about agents from the outside, this course takes you through building one, end-to-end. Across 10 lessons, you'll go from architecture overview to a working multi-agent backend with a UI, then into the part that matters most for us: testing and evaluating it properly.
By the end, you'll be able to:
- Build a GenAI agent from scratch
- Understand tool calling and MCP servers
- Run a multi-agent backend talking to an LLM
- Test and evaluate agents with confidence
You don't need to be a developer to get value from this. It's a Beginner-level course, and Jas has designed it so you can either follow along with the lecture and demos to build your understanding, or roll up your sleeves and run the code yourself. A full repository is included so you can go one level deeper whenever you're ready. Basic coding skills will help you follow the hands-on activities, and an API key for an AI tool is handy but not essential. There's a lesson on setting up an open-source model locally if you'd rather go that route.
🎓 The course is available now to Professional Members.
Curious what quality looks like when the product you're testing is an agent? This is where to begin.
