


AnrufPro is a multi-tenant AI agent platform I have built from scratch since March 2026. Its core is the agent runtime: users and agents share one principal layer, so each gets its own permissions, tools and working directory, with RBAC and ReBAC, agent-to-agent calls, a full audit trail and my own MCP servers. A workflow engine similar to n8n covers any workflow, with audit, versioning and a migration engine. Voice agents run on ElevenLabs and VAPI behind a provider-agnostic layer. More than 50 custom linters and 34 ESLint rules catch the shortcuts coding agents take: shallow assertions, hand-written mocks, skipped or flaky tests, architecture and tenant-isolation violations.
AnrufPro — AI Agent Platform
As Head of Development, built an AI agent platform from scratch, with AI-driven implementation under my direction. Users and agents share one principal layer, so each agent gets its own permissions, tools and working directory. Kept AI-written code reliable with more than 50 custom linters and 34 ESLint rules.





