I run business functions using AI agents as the execution layer.
An operator who understands both how a business works and how to make agents run it.
Most companies need to redesign their workflows for agents — but don't have anyone who understands both the business function and the AI layer. The agent operator fills that gap.
AI agent that converts conversations into sales. Handles product discovery, qualification, and purchase flow over WhatsApp. Deploys in hours, not weeks.
Interactive visualization of Composio's tool dependency graph. Maps how tools connect across the agent layer. 100+ Google integrations charted.
Personal knowledge OS built in Obsidian, managed via Claude Code and MCP. Every function documented, repeatable, and improvable.
Multi-agent systems that run business functions. Prompt stacks, tool use, MCP integrations, workflow design.
Daily user. Vault management, agent deployment, system building. Operator-level fluency.
Catching agent failures before clients do. Running evals before launch, monitoring live systems, keeping agents on spec.
Deep understanding of how businesses actually run — growth, ops, revenue. Agents work better when the person directing them knows the domain.
MCP server design, tool dependency mapping, integration architecture. Built the Composio graph from 100+ API tools.
Obsidian-based knowledge management. Every function documented, repeatable, and improvable. The operator OS.
AI agent that converts conversations into sales. Built for e-commerce operators who want revenue, not more headcount.
Visit Sella →Signal for the agent layer. Operator-focused coverage of AI infrastructure, MCP, production systems, and what actually runs.
Follow @harnessoperator →If you're building something that needs to run on agents, this is where we start.
X / Twitter @facundofranco_ LinkedIn facundo-franco Linktree facundofranco