Tamar Peretz

Tamar Peretz is the author of Andy’s AI Playbook, publishing practical resources for reliable, evidence-based AI work.

Author of Andy’s AI Playbook

I created Andy’s AI Playbook to make AI-assisted work more reliable, inspectable, and repeatable.

My work focuses on the gap between what an AI system appears to have completed and what can actually be verified: which sources it used, what context it received, whether it followed the required process, and whether its output is supported by evidence.

Tamar Peretz, author of Andy's AI Playbook

What I work on

I research and write about the practical and technical problems involved in using, reviewing, and building AI systems, including:

  • reliable AI workflows for research, writing, review, planning, and implementation
  • prompt engineering and reusable execution patterns
  • source verification, evidence boundaries, citation discipline, and factual review
  • AI agents, memory, routing, orchestration, and tool use
  • prompt injection, trust boundaries, and automation risk
  • model behavior and evaluation
  • the difference between fluent output and verified results

What I publish

Andy’s AI Playbook brings this work together in several formats:

  • Articles explain technical concepts, system behavior, risks, and design decisions.
  • Guides turn those ideas into repeatable procedures.
  • Prompts provide reusable instructions and control patterns.
  • Policies define evidence, source, review, and output requirements.
  • Reference material provides diagrams, terminology, platform maps, and technical lookup resources.

Why I built this site

AI systems can produce convincing outputs without proving that they reviewed the complete evidence, used the correct sources, followed the required workflow, or achieved the user’s actual goal.

I built this site to make those boundaries visible and to provide practical ways to control, review, and improve AI-assisted work.

The goal is not simply to produce better prompts. It is to build workflows in which claims, evidence, context, execution, and completion can be inspected.

About Andy’s AI Playbook

Andy’s AI Playbook is where I publish and maintain this work. The site is published under Andy Agent Lab.

Follow