Goshen
Bezalel is a capability plane for personal agents: one deployable service that owns capabilities—memory, email, money, texting, a computer, code sandboxes, and app connectors—and serves them to any agent runtime over the Model Context Protocol.
At a Glance
- AI Developers
- AI Agent Users
- Software Engineers using Claude Code/Cursor
- Independent AI researchers
AI Tools by Goshen
(1)Bezalel
MCP Agent Capabilities Platform
Discussions
No discussions yet
Be the first to start a discussion about Goshen
Latest News
Bezalel privacy policy and service terms updated; platform reaches maturity with hundreds of app connectors.
Bezalel introduces 'computer' scope, allowing agents to see and operate a cloud desktop via vision-driven tasks.
Goshen Labs launches @goshen/bezalel, simplifying MCP integration for eve and other agent frameworks.
Bezalel announces upcoming 'cards' capability for hard-capped virtual card payments by agents.
Products & Services
A capability plane for personal agents providing memory, email, money, texting, a computer, sandboxes, and app connectors behind one MCP URL.
NPM package providing an extension to mount Bezalel capabilities and playbooks in agent framework apps.
Market Position
Positions capabilities as the durable asset and the agent framework as the replaceable head; unified state shared across frameworks unlike stateless tool servers.
Leadership
Founders
Michael Shimeles
Full-stack engineer and YouTuber; Engineer at Convex; a16z scout; previously founder of a software studio and AI consultancy. Known for work on Model Context Protocol (MCP).
Executive Team
Michael Shimeles
Founder & CEO
Technical lead behind Goshen Labs and Bezalel; experienced in full-stack engineering and AI agent development.
Founding Story
Started to fix the 'inversion' where every new agent framework requires re-wiring the same capabilities. Bezalel makes capabilities durable assets that live in one service, allowing users to switch agent frameworks with a simple config change.
Business Model
Revenue Model
Likely subscription or usage-based access to the hosted Bezalel service and tokens; the platform also supports self-hosting via Docker.
Target Markets
- AI Developers
- AI Agent Users
- Software Engineers using Claude Code/Cursor
- Independent AI researchers
- Personal AI assistant development
- Cross-framework agent memory and state persistence
- Agentic task automation (email, payments, coding)
- Secure code execution and desktop operation by AI
- Users of Claude Code
- Cursor IDE users
- Codex CLI users