본문 바로가기

Product

Finding PMF with Chilli

728x90

Goal: Making a best ai friend chatting for a long time than Character.ai

 

1st mvp - "Chilli" Landing page for AI character creator (6.16 ~ 6.22)

- Hypothesis: creators want to make their own character and get followers

- Action: Making a landing page describes 2 values

     1. Watch how your followers role-playing with your ai character

     2. Upload contents and make new role-play everyday)

- Result: Low interests / Get no testers

- Next plan: Before verifying creators, we need to develop user's platform

 

2nd mvp - "Rollet" role-playing platform with ai character (6.23 ~ 7.6)

- Hypothesis: Users can get role-play ideas from character's feed and keep a long chat

- Action 1: Making a platform users can browse ai character's daily feed

- Action 2: Adding "creating customized character" flow and enhancing "discovery" ui layout

- Result: 33% retention / but well-made feed gets many messages

- Next plan: To get more ideas related to role-play, we need to give scenarios and missions to users rather than feed contents (= to make role-play longer, we need to provide input during chat session rather than preceding session)

 

3rd mvp - "Chatcrush" role-play (7.7 ~ )

- Hypothesis: Users can get more attachment and likeability while scenario based role-play, and then keep longer chat after clearing all the sessions

- Action 1: Making a game-like platform users can complete the scenario stages and get reward from their quality of chat

- Action 2: Adding "browsing characters" page and "chat reward" ux flow / Enhancing "scenario stage" ui layout

- Result: 11% retention / Most of the users didn’t beat the stage over 2 step. It is more difficult to keep role-playing when scenario is specific. In conclusion, we can’t find the solution with the chatbot, since it is only valuable for NSFW chat

- Next plan: Giving users mission and reward system is fun and positive experience, but it is not the key to keep long conversation