Case Study // mailstrom

MAILSTROM

AI-powered email outreach copilot with Gemini/Ollama and native desktop integration.

AppNative

AILocal+Cloud

BackendFastAPI

DatabaseSQLite

DEMO_RECORDING

[ VIDEO ASSET REQUIRED ]

02 // Architecture Execution

Built a native desktop app using Tauri (Rust) and a Python FastAPI backend. Integrated Ollama for local email classification and Gemini API for drafting highly personalized emails. Implemented APScheduler for automated interval sending via Gmail OAuth.

01 // The Problem Context

Email outreach often requires managing lists manually and using generic templates, leading to low engagement. The goal was to build a local-first copilot to automate personalized email campaigns.

System Stack

PythonFastAPITauriRustOllamaGemini

core_module.ts

# FastAPI Mailstrom Backend
@app.post('/draft')
async def draft_email(contact: Contact):
    # Local or Cloud LLM Routing
    draft = await generate_with_gemini(contact.context)
    scheduler.add_job(send_email, 'date', run_date=schedule_time, args=[draft])
    return {'status': 'Scheduled'}

Source Code