dtyq/magic

What it solves

Magic is an enterprise-grade AI agent platform designed to overcome the limitations of personal AI tools when scaled to an organization. It addresses issues such as fragmented internal data, unpredictable API costs, security risks associated with third-party tools, and the lack of human oversight for high-risk autonomous actions.

How it works

The platform provides a centralized hub where internal systems (ERP, CRM, databases) and domain expertise are encapsulated into "digital employees" or Expert Agents. Every employee is assigned a personal AI assistant that can mobilize these specialist agents and internal data to complete tasks. To ensure security, agents run in proprietary sandbox containers isolated via VPCs and private endpoints. The system includes a built-in rendering framework that converts AI text output into finished business artifacts like PPTs, Excel files, and dashboards.

Who it’s for

• Solo founders and one-person teams (OPC/OPT) who want to operate with the output of a large team by automating roles like marketing, legal, and finance.
• Enterprises and mid-size teams seeking a secure, centralized platform to manage AI agents, control costs per department, and preserve institutional knowledge.

Highlights

• Results as Deliverables: Transforms chat output into professional reports, PPTs, and data dashboards.
• Enterprise Security: Uses sandbox isolation and sidecar network proxies to ensure data privacy and tenant isolation.
• Human-in-the-Loop: Requires explicit human approval for high-risk operations (e.g., deleting data or sending emails).
• Granular Cost Control: Allows for daily budget caps set at the department, user, or agent level.
• Open Ecosystem: Compatible with Anthropic Skills and OpenClaw Skills ecosystems.