How this works

Each scenario is a character with a consistent backstory, personality, and set of boundaries, not a decision tree with pre-written responses. Every message you send is read in full and answered live, so the conversation responds to what you actually say. If you ask something different than another trainee did, you will get a different reply, though it stays consistent with the same character and situation.

That also means the exact wording is never identical twice, even in the same scenario. Two real conversations with the same person are never word-for-word the same either. Restarting a scenario also gives you a different concrete reason the character is reaching out, not the same story over again, while testing the same skill.

The character isn't locked into a fixed opinion either. Offer a genuinely thoughtful reframe or a piece of feedback that actually lands, and they can visibly soften or reconsider how they're talking about their problem, the same way a real person's mindset shifts when someone says something that actually rings true. A shallow or rushed comment won't move them, it has to genuinely make sense.

Text scenarios are a live, ongoing back-and-forth, closer to a real-time conversation. Sometimes the character reaches their own natural stopping point and says goodbye on their own, whether the conversation went well or not, a natural ending isn't a reward for a perfect conversation, it's graded honestly afterward either way. When that happens, the conversation ends right there with that closing message and moves you straight to feedback. Otherwise, you end it yourself when you are ready, but ending is part of the skill too: type a closing message before you press "End & get feedback", a real conversation should not just cut off mid-thought.

Email scenarios work the way a real peer support email actually does: you get one message from the character, and you get one reply. There is no back-and-forth, so that single reply needs to cover everything, rapport, reflection, and a safety check if the situation calls for one. As soon as you send it, the conversation ends and gets graded.

Call scenarios are their own roster of characters, built for hands-free voice practice: the character speaks out loud, you talk back, and it sends automatically once you pause, no typing and no button to press. Under the hood a call plays out exactly like a text conversation, same live back-and-forth, same natural endings, same "type a closing message before you end it" rule, just spoken instead of typed.

Scenarios involving real risk track a hidden measure of how much the character trusts you, based on how you engage. Reflect their feelings accurately, ask open and unhurried questions, and check in on their safety directly and with care, and they tend to share more. Rush to advice, ask closed or rapid-fire questions, or respond without warmth, and they tend to become guarded, sometimes harder to reach rather than easier.

This is intentionally invisible while you are in the conversation, just as it would be in a real one. You will only see the effect afterward, in your feedback report.

Scenarios are grouped into nine tiers, ordered from easiest to hardest, so no single tier requires clearing many more scenarios than another. Each tier has the same skill tested across text, call, and email, rather than splitting the written and live versions of a skill into separate tiers. Tiers 1 and 2 focus on rapport and active listening with no real risk present. Tier 3 adds checking directly on someone's safety. Tier 4 increases the emotional difficulty through reflective listening and question craft. Tier 5 introduces a different kind of challenge entirely, coaching someone who is worried about a friend rather than assessing the help-seeker themselves. Tier 6 tests holding a boundary with warmth and staying patient through hostility. Tiers 7 through 9 involve real risk disclosures: a direct statement of risk, grief and hopelessness tied to a hard diagnosis, risk hidden behind a different presenting problem, and the highest-stakes scenarios in the library.

After you end a conversation, it is graded against a ten-part rubric that covers rapport, reflective listening, question quality, non-judgmental tone, facilitative problem-solving, and, where relevant, how directly and thoroughly safety was addressed. Feedback quotes exact lines from your conversation to show what worked, plus two or three concrete priorities for next time. It is written gently and constructively, not as a list of mistakes.

Each point in your feedback also explains how that moment likely affected the help-seeker, not just whether it was technically correct. A strength shows how it likely helped them feel understood, safer, or better about themselves. A priority shows how the original moment likely felt to them and why. How you say something matters as much as what you say.

Strengths are never invented just to have something positive to say, if a conversation genuinely had no bright spot, that section stays honest and empty instead. And a formal closing summary is never required to earn a good closing score, a warm, honest sign-off is enough on its own when the conversation is genuinely wrapping up.

Each conversation starts with a points budget instead of a score you build toward. Anything the grader marks down costs some of that budget. Doing well never adds to it. Certain serious misses, like never checking on someone's safety when it clearly mattered, are flagged directly and can fail a conversation on their own, regardless of how the rest went. Pass every scenario in a tier and you earn that tier's achievement.

Every message you send is checked as it goes. A message that shames, mocks, or dismisses the character, or that crosses a real boundary, ends the conversation on the spot and fails it, regardless of how the rest of the conversation was going. Your report explains exactly what triggered it.

This is a practice space for building peer support skills, not a real crisis service. Every character here is a simulation, not a real person, and nothing you do in a conversation reaches anyone who needs help.

If you or someone you know is struggling right now, this is not the right place for that. In the US, you can call or text 988 any time to reach the Suicide & Crisis Lifeline.