CEO at Cubitrek, an AI-first agency. COO at MindKeepr, the operating memory for governed AI. Most of my time goes to one problem seen from two sides: making AI agents useful in production, and making them safe to trust with real work.

An open standard I authored for verifiable AI agent identity and authority: one signed JSON file at /.well-known/agent-passport.json, keyed in DNS, with a verifier, spend caps and signed receipts. MIT licensed. Spec and reference implementation: Cubitrek/agent-passport.

• Cubitrek builds websites, apps and AI agents, runs cloud and security, and gets brands found on Google and named in AI answers.
• MindKeepr shows what your organisation knows, what it still needs to verify, and who can close the gap, then keeps the approved answer available to authorised people and AI agents.

The public code lives in the Cubitrek organisation:

• agent-passport: verifiable identity and authority for AI agents.
• shopify-multistore-mcp: one MCP server across multiple Shopify stores, with a production write-gate.
• bazaar-commerce: multi-vendor marketplace engine for cash-on-delivery markets, built on MedusaJS and Mercur.

• X: @phyxan
• Free AI readiness check: cubitrek.com/audit