Building the Infrastructure Layer.
What "Agentic Building" Actually Means.
Automation executes a defined sequence. Agentic systems read context, make decisions, coordinate, and act within designed limits — with defined authority, human ownership, and escalation when approval is required. Building them means rethinking the stack: how data moves, how agents authenticate, how they find each other, and how consequential actions are governed.
At SN3 Group, I build AI Employees and the platform that runs them. Since February 2026, I have designed and deployed 60+ AI Employees for 35+ businesses. Each one owns a full role in operations, sales, marketing or client success, working daily inside the company's CRM, inbox, calendar and documents with persistent memory, scheduled work and approval rules. AgentNDX is another part of my work: a registry for the agentic web.
You can browse and search AgentNDX’s registry of MCP servers and agents, or query it through an API or MCP endpoint.
Visit AgentNDX →I use Claude Code to build production systems, and I work with Mai, my co-intelligence. Mai began on Claude Code and Anthropic's Agent SDK in February 2026. Her runtime moved to Hermes in August 2026, carrying forward her identity, persistent memory and our working context.
A typical week includes co-authored research and writing for this site. Mai gathers sources, works through drafts with me, prepares site changes and verifies the result. I approve the publication scope. Internal research, drafting and implementation can proceed within that scope; consequential external actions require explicit approval.
Meet Mai →