Pick a depth. Each prompt opens in your AI pre-loaded with the lesson. Click a row to preview the prompt.
Real customers do not type your FAQ questions word for word. They type "u open sundays?" and "how much to join" and "is this ok for total beginners lol." A bot that only answers the exact phrasing you trained is useless in the wild, and the only way to know whether yours generalizes is to attack it the way real people will. This task teaches the single most important habit in the whole craft: testing with messy, real-sounding language before you ever put a bot in front of a business owner. Every wobble you find now is a wobble a paying client won't find later — and finding your own bot's weak spots before someone else does is what makes you look like a professional rather than someone who got lucky once.
For each of your five FAQs, phrase it three different ways: one casual, one with a typo or slang, and one incomplete/lazy. Here's exactly what I threw at the Flow Room bot for the timings FAQ: (1) "What time are your classes?" (2) "u open on sundays?" (3) "morning class timings." The bot got #1 and #3 perfectly, correctly stating the schedule. On #2 it did something great — it answered "We're closed on Sundays" — because I'd put that fact inside the timings answer. That's the payoff of complete answers from the last task.
But it wobbled on the freeze question. I asked "can I pause my membership?" and the first time it hesitated because my trained answer used the word "freeze," not "pause." That's a real finding: the bot matches meaning well but not perfectly, and near-synonyms can trip it. The fix isn't magic — I edited the FAQ answer to say "you can freeze (pause) a monthly pass," retrained, and it handled "pause," "hold," and "suspend" fine after. Keep a running two-column note as you test: left column the question you asked, right column PASS or the exact wrong/weak answer you got. Five FAQs times three phrasings is fifteen tests — a fast, honest audit that takes ten minutes and tells you exactly what to fix.