Roadmap

Engage in conversations with multiple characters simultaneously. You will be able to select the speech order/rules, and @mention characters to have them speak.

A shared environment for characters to inhabit. Villages have a set of characters and a set of locations with a shared context. When you connect to a village for the first time in X hours, you will be greeted with a summary of events that have happened since you left, and a prompt to go interact with a specific character. For example; "Since you were here last, Liliana has opened a bakery and she's very excited. Why don't you check in on her?". Most of the characters in the village will likely want to gossip about Liliana's bakery. A village is just a construct, and can be any kind of location; share house, college campus, space station, office building, town. You have full creative freedom with your villages. You will be able to create multiple villages.

Personalized profiles for creators and users to showcase their characters, achievements, and contributions to the Mandragora community.

Sometimes you just want more information, but if you ask the character you're chatting to it derails the conversation. We will add a series of "slash commands" that will allow you to expand on the current scene in a variety of ways. "/describe the smell of the room", "/elaborate on the position we're in", "/direct towards the restaurant we discussed" "/emotion how do they feel about that?" "/shake-it-up" (no further input). After some time and testing, we'll solidify them and add a menu for choosing these as an alternative to inputting them

Characters (and users) should be able to carry things, and use them. Your inventory will be displayed in the chat UI.

A set of visual frames and backgrounds for chat messages. These will react to each message and select one of 25 different frames based on emotional content. For example, if a character (or the chatter) says something cold and standoffish, the chat bubble will appear to be covered in icycles. Similarly in a heated exchange it may appear to be on fire, or in a romantic exchange it may appear to be covered in roses.

Characters should be able to express themselves in more ways than just text. We have a plan to allow characters to use images and gifs to express themselves, and to allow users to upload their own images and gifs for their characters to use. These will be optionally provided by character creators for 25 different emotions, and will be switched between automatically as the chat evolves.

Sometimes while you're chatting with a character, you meet someone new. Mandragora will automatically extract new characters into profile stubs, and update them as you continue to chat. NPCs will have the option to be promoted to a full character profile via the character builder.

A more advanced form of character expression. We want to bring your characters to life with dynamic, animated portraits that react to the conversation in real-time. We'll use a one-time generation at character build time to transform static character images into responsive, emotive faces that blink, smile, and express a range of emotions, and then we'll interpolate between these based on the context of your chat. This will be limited to close up character portraits, and will eventually be used for secondary characters or NPCs

A more advanced form of the live portraits feature. Some time ago we rigged together a prototype that allowed us to fully animate a character from a static image. We have a plan to bring this feature to Mandragora. Your character image will have to conform to a specific format, but once active, they will be able to breath, move, blink, act shocked, happy, sad, and more, depending on the evolving context of the chat. These animations are generated on the fly.

Mandragora already does multiple layers of inference behind the scenes during your chats. We will add the ability to generate an image based on a current description of the scene. Including the present character(s) when possible.

Speaking to your character companions and freinds should feel natural and normal. We have a working prototype in the lab for this.

A lot of text to speech implementations are so bad that they may as well not have been shipped. We want to buck that trend and use the best available text to speech for our users. We're still in the R&D phase with this one, but our early tests are already promising. The secret is using a TTS stack that can inhale.

Mandragora will adapt the background image of your chat, and add dynamic particles to reflect scene conditions; such as falling leaves or snow.

We have a lot more planned - Mandragora has been in ideation and R&D for years, but for now; we want to make sure we can continue to deliver on our promises before we start talking about the next big thing. If you want to chat about Mandragora, or have any ideas, please join our Discord server.