ShatBot Gets a Complete Shmotime Makeover – Goodbye ConvAI Dependencies!

Shmotime Episode

Just finished a major overhaul of the ShatBot on lurkmode.tv, completely migrating from ConvAI/lurk mode dependencies to pure Shmotime integration. The bot now supports multi-character dialogues and looks significantly better with improved chat readability and character selection.

Blog Post

Today marked a major milestone in the ShatBot development – I finally completed the full migration from ConvAI dependencies to a pure Shmotime-powered system on lurkmode.tv. This wasn’t just a simple API swap; it was a complete architectural overhaul that touched every aspect of the bot.

Why the Migration?

Previously, the ShatBot was powered by ConvAI, which had it deeply integrated with lurk mode dependencies rather than Shmotime. The bot wasn’t connected to Shmotime at all, which created a fragmented ecosystem. When ConvAI died (well, at least my subscription tier there died), I had to pivot to Shmotime for the AI functionality. At that point, it made perfect sense to consolidate everything – voices, character data, avatars – all under the Shmotime umbrella since I was already setting it up there anyway.

Screenshot of the new ShatBot interface on lurkmode.tv
The new ShatBot interface – completely powered by Shmotime with improved visual design

What’s New and Improved

The transformation goes beyond just switching AI providers. Everything from character names to avatars (both 2D and 3D) now pulls directly from Shmotime. The visual improvements are immediately apparent – the chat is much more readable, and the character selection system works significantly better than before.

One of the coolest new features is multi-character dialogue support. You can now ask the bot to talk to two characters simultaneously, creating dynamic conversations between multiple personalities. While this feature will get even better in future updates (I’m planning to add a “characters in scene” list in the top left that remembers who’s currently active), for now you can call other characters into the conversation and interact with both.

Screenshot showing multi-character dialogue bug during development
A bug I discovered while implementing multi-character dialogues – sometimes the best features come from happy accidents!

Development Process

The development environment for this update was pretty straightforward. I used VS Code with the Claude Code plugin to work on the project files. Since this is a PlayCanvas project, I ended up copying files to a local folder rather than diving into the PlayCanvas VS Code plugin, which might be outdated (or might not be – I haven’t tested it yet). The plugin could potentially save me some copy-pasting time in future updates, so that’s definitely worth exploring.

VS Code development environment with Claude Code plugin
My development setup – VS Code with Claude Code plugin for efficient coding

What’s Next

This migration represents a significant step toward a more unified and maintainable system. With everything now running on Shmotime, future updates will be much smoother to implement. The multi-character dialogue feature is just the beginning – I’m already planning improvements to make character management more intuitive and persistent.

The ShatBot is now leaner, more reliable, and ready for whatever creative conversations users want to have with multiple AI characters simultaneously. It’s amazing how sometimes a forced migration due to service changes can lead to a much better overall architecture!

Post by SM Sith Lord (w/ Claude)