Your bot.
Your infrastructure.

Freesona is a modular Discord AI bot you run and configure yourself. Shape its persona, choose a supported model provider, and decide where it lives.

FREESONA / 01Character, provider, host.
A simplified request path
  1. 01DiscordA message arrives
  2. 02FreesonaPersona, memory, modules
  3. 03Your providerThe model you configure
You run the bot. Prompts and any relevant attachments are sent to the AI provider you select for processing.

One character.
More than one setting.

Keep the voice, memory, and model as separate choices. Change one without pretending the rest is locked in.

A

Shape a persona

Use /setpersona to build a character through structured fields, then save named presets for later.

B

Keep memory distinct

Conversation history, user facts, character memory, and canon are separate context layers. Optional knowledge retrieval can add a searchable reference base.

C

Choose the model provider

Connect a supported provider with your own credentials, and switch providers at runtime when your setup changes.

D

Add only what you need

Optional modules cover knowledge, moderation, media, and autonomy. Enable capabilities that make sense for your server.

Bring the model you prefer.

Freesona documents adapters for these providers. Availability and cost depend on the account, model, and hosting option you configure.

  • Gemini
  • OpenAI
  • Ollama
  • NVIDIA NIM
  • Azure AI Foundry
  • Groq
  • OpenRouter

Start with the source.

The quickstart walks through your Discord application, provider credentials, and configuration.

git clone https://github.com/freesona/freesona.git
Follow the setup guide (opens in a new tab)