A large language model, or LLM, is an artificial intelligence trained on very large amounts of text to understand a request and write an answer. It is the "brain" of AI assistants and AI agents.
Hosted models and open models
- Hosted: you use them at their publisher, through an application or an API (Claude, GPT, Gemini…).
- Open: their weights are published and you can install them yourself (Llama, Mistral, Qwen, DeepSeek, Gemma…). Companies that want to keep their data in-house often choose this self-hosted AI.
What a model cannot do on its own
- A model writes, summarizes, translates, reasons. But it has no access to your contacts, your statistics or your sending domain.
- To act, it needs tools: that is the role of MCP servers, which the model uses through an MCP client.
Language models and Mailpro
- Mailpro offers built-in AI tools, such as the email writer.
- And with the Mailpro MCP server, the model of your choice, hosted or installed in-house, can prepare and send your emails and SMS from your account.
Mailpro and AI: your model, your account, your sends. Start for free