The Hard Part is conversation practice for the talks that keep you up the night before. Roleplay difficult conversations with an AI — firing someone, hard feedback, saying no — at three difficulty levels, by voice or text, and get a scorecard after.
Pick a situation, read a thirty-second briefing, then talk it through with a character who pushes back the way a real person would. Speak out loud or type. About five minutes a go — a conversation ends after twelve exchanges, or whenever you tap Wrap up.
Then you get a scorecard: a mark out of five for Clarity, Empathy, Firmness and Outcome, two moments quoted back from your own words where you did well, and one concrete thing to try next time. One thing, not five.
THIRTEEN SITUATIONS
Each one trains a different kind of difficult. The person who talks over you. The one who never argues and somehow leaves with your yes. The one who goes quiet. The one who has a reason for everything. Getting good at one does not make you good at the others. Five of the thirteen are built for people new to managing others — feedback to the peer who trained you, a late-night boundary, saying no to your own boss, a first underperformance conversation, telling a friend she didn't get the role.
THREE DIFFICULTY LEVELS
Warm-up to find the words. Realistic for a fair fight. Hard Mode when you want to be tested. The person does not change — they just stop making it easy.
VOICE OR TEXT
Say it out loud and hear it answered, or type instead. Your speech is turned into text by your phone's own speech recognition, so the audio never leaves your device.
WHAT THIS IS NOT
This is practice, not HR or legal advice. Every character is invented. Please do not enter real colleagues' names or identifying details. If a conversation turns towards self-harm, abuse or crisis, the session stops, shows support information, and is not scored.
FREE AND PRO
Free: two rotating scenarios each week — one from the original eight, one from the new-manager five — three sessions a month, text mode. Free sessions carry ads — a short ad once a session ends, and an optional one you can choose to watch to keep talking when a session reaches its exchange limit.
Pro, a subscription: all thirteen scenarios, Hard Mode, voice practice, your session history, and no ads.
The characters' replies are generated by an AI model. Press and hold any reply to report it, without leaving the app.
Practice difficult conversations with AI: firing, feedback, saying no. Scored.