Megan AI
A downloadable game for Windows
About This Game
Overview
Megan AI is a single-player sandbox that turns your PC into a fully offline AI companion.
Load any compatible GGUF language model, give it a voice and face, and chat by text or microphone while a MetaHuman lip-syncs every reply.
Key Features
- Local language models – Four Apache-2.0 7 B models included; drop in your own GGUF files.
- Voice input & output – Speech-to-text via Wit.ai* and crystal-clear TTS with SherpaOnnx.
- Real-time MetaHuman – High-fidelity avatar driven by UE 5.6 with precision lip-sync.
- Prompt sandbox – Edit system prompts, adjust temperature, context window, GPU layers.
- No cloud, no telemetry – All conversations stay on your machine.
- Early Access roadmap – Weekly updates: model hot-swap, Workshop personality sharing, GPU off-load optimisations.
*Voice input is optional. No audio is stored or uploaded.
Why Early Access?
We’re a tiny indie team, and your feedback will decide which models, prompts, and features we polish first. Join the community hub, post conversation screenshots, and help us shape the future of personal AI!
AI Generated Content Disclosure
The developers describe how their game uses AI Generated Content like this:
Megan AI runs a local language model on your PC to generate dialogue and text-to-speech replies in real time. No conversation data leaves your machine.
Mature Content Description
The developers describe the content like this:
Megan AI is a free-form conversational experience. All dialogue is generated live by an on-device language model based on your prompts. While we ship a default “no explicit content” system prompt, the AI may still produce occasional coarse language, references to mature themes, or user-requested role-play that some players find inappropriate. There is no built-in graphic violence, pornography, or hate speech, and no visual depictions of nudity or gore. You can clear the chat history or reset the AI at any time
System Requirements
- Minimum:
- OS: Windows 10 64-bit
- Processor: Intel Core i5-8400 / AMD Ryzen 5 2600
- Memory: 16 GB RAM
- Graphics: NVIDIA GTX 1060 6 GB or equivalent
- DirectX: Version 11
- Network: Broadband Internet connection
- Storage: 15 GB available space
- Sound Card: DirectX-compatible
- Additional Notes: Uses on-device AI; lower-end CPUs/GPUs may see slower replies.
- Recommended:
- OS: Windows 10/11 64-bit
- Processor: Intel Core i7-10700 / AMD Ryzen 7 3700X
- Memory: 32 GB RAM
- Graphics: NVIDIA RTX 3060 12 GB or better
- DirectX: Version 12
- Network: Broadband Internet connection
- Storage: 20 GB available space
- Sound Card: DirectX-compatible
- Additional Notes: GPU off-load improves AI speed; headset mic recommended for voice chat.
Status | In development |
Platforms | Windows |
Rating | Rated 5.0 out of 5 stars (1 total ratings) |
Author | ZavGaming |
Genre | Simulation |
Made with | Unreal Engine |
Tags | artificial-intelligence, metahuman, Sandbox, Unreal Engine |
Average session | A few seconds |
Links | Steam |
Download
Click download now to get access to the following files:
Development log
- New Update 0.1.251 days ago
Comments
Log in with itch.io to leave a comment.
This was fun!
I'm wondering if it's possible to get more voices?
Like I downloaded a custom GGUF file and was able to load that.
Are the voices options tied to that model, or can I download voices separately and pair them with any GGUF model?
Thanks.
nice
if using unrestricted be aware it is a llama descendant so use the llama chat template!
Make sure you set the system prompt, oh man so fun!