Fabian H.

I help teams find out why their AI agents fail, connect Claude to business tools, and build automations they can test and maintain. If your team keeps repeating setup work or cannot explain the API bill, start with a focused review and a clear next step. What you can hire me for - AI agent audits: findings on failures, costs, permissions, and the changes worth prioritizing, backed by evidence. - Claude Code setup: repository instructions, permissions, and a repeatable task your team can run and verify. - MCP integrations: a defined connection to your API, with access rules, error handling, and documentation. - Document automation: n8n workflows that combine LLM processing, validation, and human review. My work includes AI fleet auditing, TimbraBot for financial document processing and invoice-to-bank reconciliation, and an internal Telegram agent system with approval gates and audit logs. I've delivered technology projects since 2009. Through WASD Studio, my earlier work included immersive training for Coca-Cola FEMSA and Alsea. I've also completed seven Anthropic courses, including Claude Code in Action and Introduction to Model Context Protocol. We agree on the deliverable, acceptance checks, and scope before work starts. You receive documentation and a handoff so your team can operate what we build. Based in Mexico City. English and Spanish, with overlap across US business hours. Message me on Upwork with your workflow and the issue you want to solve. I'll recommend a practical first step.