DONATE
Operational Governance for Live AI Systems — FairByDesign

FairByDesign Books & Playbooks

Operational Governance for Live AI Systems

Two practitioner-grade systems for teams that need to govern AI that is already in production — from workflow-level oversight to agent deployment in regulated environments.

These are not principles documents. They are operating manuals.

FairByDesign publishes practical governance systems for teams dealing with the real problem: AI already influencing live workflows, regulated decisions, customer outcomes, internal operations, and increasingly, autonomous agent behavior.

Start with Operational AI Readiness if you need to make a live AI workflow legible, bounded, reviewable, and defensible. Move to the Enterprise AI Agent Deployment Playbook if you are deploying action-taking AI agents and need release gates, evidence requirements, adversarial resilience, and reviewer-verifiable accountability.

Book

Operational AI Readiness

A Practitioner Guide to Governing AI That Is Already Live

By Viktoria Bakos | Published by FairByDesign

You inherited an AI governance problem that is already live. This is your operating manual.

Built for teams using AI in hiring, payment approvals, customer disputes, security triage, vendor review, and other consequential workflows. This guide shows how to inventory live systems, classify decision influence, build approval gates, preserve an evidence spine, and implement human oversight that survives contact with real work.

Includes:

Twelve chapters. Five real-world scenarios. A complete 30-day readiness sprint. Sample policy rules in YAML. Evidence logging schema. Human-in-the-loop workflow specification. Vendor due-diligence question bank. Minimal readiness test suite.

Best for: Security engineers, product managers, compliance leads, internal auditors, and platform architects who need to govern one live workflow properly.

Featured

Playbook

Enterprise AI Agent Deployment Playbook

Deploying AI Agents in Regulated Environments

By Viktoria Bakos | Published by FairByDesign

You’re deploying autonomous AI into systems designed for human operators. This is your governance manual.

For organizations building or buying AI agents that make tool calls, write to systems, trigger workflows, modify records, or act without asking permission first. This playbook translates the EU AI Act, NIST AI RMF, GDPR, DORA, NIST SP 800-53, SR 11-7, OWASP, and MITRE ATLAS into deployable controls with release gates, evidence requirements, and named oversight authority.

Includes:

Twenty-five chapters across nine parts in the Standard Edition. Two deployment scenarios. A four-gate release framework. Five-layer control stack. Human System Owner model. Threat analysis for prompt injection, tool misuse, memory poisoning, RAG compromise, and supply chain failures.

Second Edition adds:

Eight additional chapters. EU AI Act conformity assessment walkthrough. GPAI obligations. Annex III deployment guidance. Adversarial testing playbook with payload examples. MITRE ATLAS operational mapping. OpenTelemetry implementation for observability. EAAD training infrastructure and practitioner tools.

Best for: Security architects, AI engineers, privacy officers, CISOs, and compliance leads governing high-stakes agent deployments across the enterprise.

Choose the system that matches the problem

One is for governing consequential AI workflows already in production. The other is for deploying action-capable AI agents in regulated environments.

Feature Operational AI Readiness EAAD Playbook
Primary focus Governing live AI workflows Deploying action-capable AI agents
Best use case One workflow, one control problem Enterprise agent deployment program
Human oversight model Workflow-level review and approval gates Human System Owner plus release gates
Evidence and traceability Evidence spine and logging schema Adversarial audit-grade traceability
Security focus Operational AI workflow risk Prompt injection, tool misuse, memory, MCP, ATLAS
Advanced team infrastructure No Yes — training, red teaming, conformity assessment, observability

Which one should you start with?

If your problem is that AI is already shaping decisions inside a live workflow and you need a practical governance foundation, start with Operational AI Readiness.

If your problem is that autonomous or semi-autonomous AI agents are taking actions across systems and you need release gates, action controls, adversarial testing, audit evidence, and deployment discipline, start with the Enterprise AI Agent Deployment Playbook.

Both are built for people who need more than AI principles pages. They are designed for teams that need controls that survive pressure, incidents, and scrutiny.

Published by FairByDesign

FairByDesign is a public-interest media and learning platform focused on AI ethics and literacy, cybersecurity, and privacy.