Let’s demonstrate the engine by creating a new campaign in the Campaigns tab. I will create an example based on the Curse of the Azure Bonds module, but I’m free to let the system define a campaign instead. I’ll also let the engine generate an appropriate banner image:

With the campaign generated, I will go to the Play area to tell the model to set up an adventuring party. I can be as hands-on or hands-off during creation as I wish:

This screen has a number of options allowing me to customize my experience:

As the system generates the members of my party, I can review them in the party tab:

Clicking on party members shows a detailed character sheet with ability scores, inventory, proficiencies, backstory and so on:

Once all party members have been created, the sidebar in the play area shows their portraits and summary information:

We start the adventure with a detailed introduction text:

As a matter of personal preference, I tend to pick a single character to play and let the model handle the others. By default, the system expects text-based input. Note the small question mark and play icon at the end of the response.
- Pressing play narrates the latest entry using text-to-speech. I have the option of choosing custom voices per character.
- The question mark shows an indication of the cost of this interaction, based on API tokens. Since I’m using a subscription, I don’t pay anything beyond a negligible amount for text-to-speech and image generation.

While I prefer to play campaigns on a laptop using the keyboard, the engine also supports a Light mode, where the system provides a number of possible actions to take at the end of each response. This is convenient when playing on a mobile phone or tablet where text entry is a bit more awkward:

As we play, the engine builds up a detailed map of areas such as cities visited, with locations in each area for points of interest:

Active and completed quests are tracked in a separate tab:

And an index of characters along with detailed notes is also accumulated:

This information is stored on the file-system, allowing campaigns to grow indefinitely without eating up the context window.
The engine can generate images at key moments which are persisted in the gallery:

That’s the full loop — campaign creation, party generation, play, and the long-tail of locations, quests, NPCs and imagery that accumulates as you go. Future posts will dig into specific subsystems (combat, travel days, faction clocks) in more detail.