A privacy-first multi-agent application platform for e-commerce, featuring low-code LangGraph workflow authoring and cross-network distributed agent deployment.
eCan.ai (E-Commerce Agent Network) is an AI-native multi-agent platform focused on end-to-end e-commerce automation. It provides preset agent roles for marketing, after-sales support, operations, procurement & logistics, and targeted advertising. Agents communicate via the A2A protocol and support distributed deployment across LAN/WAN with mobile devices serving as remote monitors and commanders.
The core workflow engine is built on LangGraph with a graphical Flowgram editor supporting drag-and-drop node construction, run/pause/resume/step-debug, breakpoint setting, runtime state inspection, and bidirectional translation between Flowgram and LangGraph code. Browser automation integrates Browser-use, Crawl4ai, Selenium/WebDriver, and AdsPower fingerprint browsers, with a built-in OCR service for CV-guided screen understanding.
AI capabilities include MCP tool integration, built-in RAG for product and policy document retrieval, Ollama local LLM deployment, and extensible data filters for enterprise-grade privacy protection. Remote interaction is available through WhatsApp, Telegram, Slack, and other channels (Pro plan).
The technical architecture uses a Python backend + React/TypeScript frontend with a GraphQL API layer, supporting desktop applications and Docker containerized deployment with fast/dev/production build modes. Primary language distribution: Python 67.4%, TypeScript 26.5%, JavaScript 4.8%. Latest version v0.8.24 with 6,756 commits.
Pricing tiers: Starter (free, 5 daily tasks), Routine ($29/month, 10 daily tasks), By Results (performance-based, customized).
Unconfirmed information: v0.8.24 release date listed as March 16, 2026 (possibly a typo); Discord server link not publicly available; API and CLI reference docs both marked "Coming soon"; Linux installer not provided despite claimed support; Browsebase integration marked "to be added".