How Ploutos Works
One brain. Every platform.
Ploutos isn't a Discord bot with a website bolted on. It's one 8-model AI brain that runs everywhere your community lives — same mind, every surface.
Your Website
A chat widget that sits on your site and answers visitors with the full Ploutos brain — same models, same knowledge, same speed. No separate chatbot to train.
Try it: the widget at the top of ploutosai.app is the real thing. ● Responding
Discord
Ploutos lives in your server as a member of the community — answering questions, welcoming newcomers, handling the repetitive work so your mods don't have to. Mention-based, no slash commands to memorize.
See it working in the Ploutos AI server. ● Responding
Multiple Personas
One brain, different personalities per server — each deployment gets its own voice, knowledge, and rules while sharing the same 8-model engine underneath.
API
Every capability Ploutos has in Discord and on the web is available through the API — plug the brain into your own app, dashboard, or workflow.
Endpoint: worker.ploutosai.app/api/test-chat
The brain underneath
8 models, one mind
Every request — whether it comes from your website, Discord, or the API — runs through the same cascade: four free models first (Llama 3.1 8B → Mistral Small 24B → DeepSeek R1 32B → Llama 3.3 70B), then paid fallbacks (DeepSeek Flash → GPT-4o-mini → Claude Sonnet → Gemini Flash) only if needed.
One shared BrainState tracks the conversation. The platform changes; the intelligence doesn't.