We built ChaosBench around a specific assumption: that whoever sits down with it already speaks the vocabulary. API Gateway, message queue, eventual consistency — the product hands you those words and expects you to already know what to do with them.
That's a fine assumption for someone brushing up before a Senior-round interview. It's a bad one for someone who's never designed a system before and genuinely wants to know how the pieces fit together.
The gap wasn't in the difficulty — it was in the vocabulary
We didn't need a separate, dumbed-down version of the product. The actual interview loop — clarify, architect, defend — is the right thing to practice regardless of experience level. What was missing was much narrower: nothing in the product ever explained what a component was before asking you to place it, and nothing in the AI's own questions or critique ever paused to define a term before using it.
A candidate who already knows what a message queue is doesn't notice this gap at all. A candidate who doesn't hits a wall on literally the first component in the sidebar.
What we shipped
Two small, independent things, deliberately not one big "beginner mode":
Hover and tap explainers, always on, on every component in the palette — what it is and when you'd actually reach for it, in plain language, not a formal definition. This has zero cost when you're not using it: it's a tooltip, not a mode you have to opt into.
Explain Mode, a toggle in the canvas header. Turn it on and the AI defines jargon inline the first time it uses a term — in a clause, in parentheses, the way you'd explain it to a smart friend outside the field — instead of assuming you already have the vocabulary. The questions and the critique don't get any softer. The interview stays a real interview; it just stops assuming you already know the words for things.
The thing we were careful not to build
It would have been easy to build a version of Explain Mode that also generates the architecture for you from a plain-language description of what you want. We didn't, on purpose. Translating a vague requirement into an actual component graph — deciding what an "API Gateway" is for, not just what it's called — is arguably the core skill this product exists to build. Handing that translation step to the AI wouldn't have made the product more accessible; it would have quietly removed the part that matters most, for exactly the people who need to practice it the most.
Explain Mode lowers the vocabulary bar. It doesn't lower the bar on the actual thinking.
If jargon has ever been the thing standing between you and starting, turn it on and see the difference on your next session.