Mailbot uses open source models hosted exclusively on servers in Switzerland. By design, no memory is retained — every task starts fresh and ends clean. Your data never lingers on our servers.
All AI models used by Mailbot are open source, ensuring transparency, auditability, and independence from proprietary cloud providers. We select the best-performing model for each specific function.
Every request is processed in isolation. Models do not retain information between tasks. Once a job is complete, all temporary files, context, and generated content are permanently deleted.
Every email follows this secure, stateless workflow
Your email arrives at the Mailbot server in Switzerland
The sender address is checked against the allowed list. If not authorized, the email is immediately deleted.
The subject line determines which function to execute
The optimal AI model is chosen based on the specific function
The result is generated and sent back to your configured email address
All files, emails, and temporary data are permanently deleted
Each function uses the best available model for optimal results
| Function | Current Model | Data Protection |
|---|---|---|
|
fill_word_template
Fill Word templates with dynamic content |
Llama 3.3 70B | High |
|
summarize_email
Create concise email summaries |
Qwen 3 | High |
|
sum_up_email
Aggregate multiple emails into a digest |
Qwen 3 | High |
|
social_posts
Generate social media content |
OpenAI GPT-5 | Medium |
|
marketing_plan
Create comprehensive marketing strategies |
Mistral 3 | High |
|
prepare_meeting
Prepare meeting agendas and briefings |
Qwen 3 | High |
|
expenses
Process and categorize expense reports |
Gemma 3n | High |
|
phone_call
Fulfill tasks via phone (schedule appointments, inquiries, etc.) |
Deepgram Nova 2 / OpenAI gpt-4.5-mini | Medium |
|
generate_pictures
Generate images based on text prompts |
OpenAI Image 1.5 / Google NanoBanana Pro | Medium |
* Models are continuously evaluated and updated to ensure optimal performance. Your administrator cannot change these selections.
Your data security is our highest priority. Mailbot's architecture is designed from the ground up to ensure that sensitive information never persists and remains under your complete control.
After every task, all processed data is permanently deleted. This includes the original email, any attachments, generated documents, and all intermediate files. Nothing remains on our servers — your data exists only for the duration of the task execution.
Employees cannot modify or override the AI model selection. All model configurations are centrally managed to ensure consistent quality, security compliance, and adherence to your organization's policies. This eliminates the risk of unauthorized model changes that could compromise data handling.
Your firm-specific language preferences, templates, tone of voice, and formatting rules are enforced on every request. These settings are locked at the administrative level, ensuring brand consistency and compliance across all generated content, regardless of who sends the request.
Mailbot only accepts emails from pre-authorized sender addresses and only delivers results to pre-configured recipient addresses. This closed-loop system prevents unauthorized access and ensures that sensitive outputs never reach unintended destinations.
With Mailbot, you get the power of advanced AI without compromising on data security. Every interaction is ephemeral, every configuration is locked, and every email pathway is strictly controlled. This is enterprise-grade AI automation built for organizations that refuse to compromise on privacy.