Antony Zaki

Antony is a consultant at Deloitte Consulting who leads the design and delivery of Claude-powered AI solutions for finance transformation. His work spans the full arc from client-facing proof-of-concept engagements to internal platform development, with a consistent focus on making agentic AI production-ready rather than merely demonstrable. On a recent POC with a large semiconductor manufacturer, he scoped and delivered workflow-based use cases, arguing for eval-verifiable outcomes over speculative predictive work within the POC window—a discipline that has become a hallmark of his delivery approach. He has also run a Claude enablement program for a global consumer technology company, translating platform capabilities into concrete, role-specific adoption.

He technical center of gravity is evaluation-driven development for AI systems. He architects eval harnesses that give clients confidence in AI outputs, including a three-tier framework for a CFO letter drafting system spanning deterministic fact accuracy, AI-judged draft quality, and pipeline integrity—complete with a calibrated answer key for tuning the AI judge. He is currently building out a plugin ecosystem on Claude Cowork, including an ideation capability covering intake, RICE-based prioritization, program ranking, and PRD generation that feeds directly into downstream build and evaluation tooling. His broader portfolio reflects deep, hands-on fluency with agentic architectures: agent harnesses, managed agents, sandboxing, and loop-based systems, all anchored to real FP&A and CFO-level reporting workflows.

Beyond delivery, he invests heavily in enablement and governance. He runs weekly Claude office hours—supported by a structured async intake system so questions and follow-ups scale beyond the live session—and has authored AI adoption and governance strategy covering phased team enablement, proactive and reactive eval frameworks, and secure MCP server hosting through enterprise identity manage.