# doushi.ai

> The AI for Machine Learning. Turn datasets and plain-English prompts into production AI models and hosted REST APIs in minutes.

## Overview
doushi.ai (https://doushi.ai) enables software engineers, product managers, data analysts, and founders to train, test, and deploy production machine learning models with a single prompt—no ML PhD or infrastructure setup required.

## Key Capabilities

- **Prompt + Dataset ➔ Production Model**: Upload a tabular dataset (CSV, Excel, Parquet, JSON) and describe what you want to predict in natural language. doushi.ai autonomously handles data cleaning, feature engineering, and model selection in under 60 seconds.
- **Chat to Explain & Inquire**: Ask the conversational AI agent questions about model accuracy and metrics in plain English, understand prediction drivers, and generate test predictions directly within the chat.
- **Interactive Live Simulator**: Test model predictions in real time directly in the browser using dynamic sliders and inputs before writing any backend integration code.
- **1-Click Hosted Production REST API**: Deploy selected models as fast, hosted REST microservices (average 14ms latency) with secure API keys and ready-to-copy code snippets for Python, Node.js/TypeScript, and cURL.
- **Official Python CLI & 100% Exportable Code**: Install via PyPI (`pip install doushi-cli`) or run via `pipx run doushi demo`. Train, predict, inspect logs, and download complete, transparent Python code with zero vendor lock-in.

## Security & Compliance

- **German & EU Data Residency**: All compute instances, storage, and sandboxes are hosted in ISO 27001 certified data centers in Frankfurt am Main, Germany (EU).
- **GDPR & DSGVO Compliant**: Strict adherence to the EU General Data Protection Regulation and the German Federal Data Protection Act (BDSG).
- **Data Privacy & Ownership**: Customers retain 100% ownership of their datasets, models, and predictions. Customer data is strictly isolated and never used to train foundation models.
- **Sandboxed Execution**: Model training and evaluation run in secure Docker container sandboxes with isolated memory and execution boundaries.
- **Encryption**: TLS 1.3 in transit and AES-256 at rest.

## Pricing Plans

- **Free ($0/mo)**: 1 Project, basic model generation, live interactive simulator, community support.
- **Starter ($29/mo)**: 5 Projects, automated feature engineering, standard hosted REST API endpoints.
- **Professional ($79/mo)**: Unlimited projects, fast training, batch predictions, priority email support.
- **Growth ($199/mo)**: Dedicated high-throughput endpoints, sub-20ms inference latency, team collaboration, priority support.
- **Enterprise (Custom)**: Custom SLA (99.99%), dedicated VPC / on-premises deployment, dedicated GPUs, SSO/SAML, custom security review & DPA.

## Links & Endpoints

- Website: https://doushi.ai
- Web Application: https://app.doushi.ai
- GitHub Organization: https://github.com/doushi-ai
- Official CLI (GitHub): https://github.com/doushi-ai/doushi-cli
- PyPI Package: https://pypi.org/project/doushi-cli/
- Privacy & Security: https://doushi.ai/privacy
- Support: mailto:support@doushi.ai