AI works in operations when it is treated like any other change to a process: start from the work, standardize first, decide who checks the output, measure the result, and control the risk. The same habits that run a good quality system, such as baselines, sampling, FMEA, change control, and audits, apply directly to AI.
This hub is about running AI, not building it. It covers finding and choosing use cases, building a business case, governing AI risk, designing the human role, and managing agents that act with some autonomy, and it ties AI to the improvement methods and hubs elsewhere on the site. Start with the cycle below, or jump straight to the tool you need.
About This Hub
Suggested Learning Path
- Where AI Fits in Operations — the kinds of AI, and whether to automate, augment, or keep a task with a person
- AI Use-Case Selection and Business Case — rank ideas and build a realistic case
- AI Use-Case Prioritization Matrix — score and rank candidates with editable weights
- AI Business Case and ROI Calculator — payback, cash, and sensitivity
- AI Governance and Risk for Operations — frameworks, risk tiers, FMEA for AI, and a policy
- Human-in-the-Loop Process Design — autonomy levels, review design, and sampling
- Agent Autonomy and Oversight Planner — the autonomy level the risk supports, and its controls
- AI Agents in Operations — design, test, and deploy an agent in stages
The AI Operations Cycle
The cycle repeats: what you learn from operating and measuring feeds the next round of finding and choosing. People and change sit at the center because every stage depends on the people who do and check the work.
| Stage | What it does | Guides | Tools | Templates |
|---|---|---|---|---|
| 1. Find | Map the work, remove waste, break it into tasks, and rank the AI ideas. | Where AI Fits in Operations; Use-Case Selection and Business Case | AI Use-Case Prioritization Matrix | AI Use-Case Register |
| 2. Design | Decide the human role, the autonomy level, and the checks built into the flow. | Human-in-the-Loop Process Design; Mistake-Proofing | Agent Autonomy and Oversight Planner | — |
| 3. Govern | Set policy, sort uses by risk, assess failure modes, and name owners. | AI Governance and Risk for Operations | RACI Matrix Builder | AI Risk Register; AI Acceptable-Use Policy and Tool Register |
| 4. Pilot and validate | Test on real cases in shadow mode, with exact intervals and a stop rule. | Measurement System Analysis; AQL Sampling | Sample Size and Confidence Calculator | — |
| 5. Operate | Deploy in stages, with permissions, limits, monitoring, and change control. | AI Agents in Operations; Management of Change | — | AI Agent Charter and Runbook |
| 6. Measure and improve | Compare results with the baseline and the case, and feed back what was learned. | Use-Case Selection and Business Case; SPC Control Charts | AI Business Case and ROI Calculator | — |
| People and change | Involve the people whose work changes, and build the skills to check and use AI well. | Change Management for Improvement; Leadership Principles | — | Skills and Training Matrix |
The hub is growing in phases. Stages 4 and 6 draw mostly on existing quality guides and tools until more AI-specific pages are added.
AI Operations Guides
Where AI Fits in Operations
The kinds of AI, automate versus augment versus keep, Lean first, and a worked complaint-handling example.
AI Use-Case Selection and Business Case
Six scoring criteria with gates, the business case, and a worked payback and sensitivity example.
AI Governance and Risk for Operations
NIST AI RMF, ISO/IEC 42001, the EU AI Act, a governance structure, and a worked FMEA for AI.
Human-in-the-Loop Process Design
Five autonomy levels, review design, automation bias, error-proofing, and sampling plans.
AI Agents in Operations
Designing an agent like a job, permissions and guardrails, shadow testing, and staged deployment.
AI in Manufacturing Quality
A practitioner field guide to AI in quality.
Using AI to Perform Root Cause Analysis
AI-assisted root cause analysis in manufacturing.
Change Management for Improvement
Bring people along when work changes.
Process Safety and Management of Change
Change control that also fits models, prompts, and tools.
Mistake-Proofing
Design the error out of the process.
Measurement System Analysis
Check that checkers agree, including AI and its reviewers.
AQL Sampling
Sampling plans for checking what AI lets through.
Tools and Calculators
AI Use-Case Prioritization Matrix
Score and rank AI ideas on six weighted criteria with gates and tiers.
AI Business Case and ROI Calculator
Payback, 36-month cash, and sensitivity for one use case.
Agent Autonomy and Oversight Planner
The highest autonomy level the risk supports, and the controls it needs.
Project Prioritization Matrix
Rank improvement projects.
Impact-Effort Matrix Builder
Sort ideas by impact and effort.
Risk Matrix Builder
Plot risks by likelihood and impact.
RACI Matrix Builder
Clarify who does what in AI governance.
Sample Size and Confidence Calculator
How many outputs to check.
Templates
AI Use-Case Register
Score and track AI ideas from idea to live use.
AI Risk Register (FMEA for AI)
Rate AI failure modes before and after controls.
AI Agent Charter and Runbook
Purpose, permissions, guardrails, runbook, KPIs, and change log.
AI Acceptable-Use Policy and Tool Register
Starter policy, data classes, a tool register, and acknowledgements.
PFMEA Template
Failure mode analysis for a process.
Skills and Training Matrix
Track the skills AI changes.
AI Terms You Will Meet
| Term or framework | What it means for improvement work |
|---|---|
| AI system | A system that, for a set of objectives, produces outputs such as predictions, content, or decisions that influence its environment (after the OECD definition). |
| Generative AI | AI that produces new content, such as text, images, or code, from a prompt. |
| Large language model (LLM) | A model trained on large amounts of text that can read, write, and answer in natural language. |
| AI agent | A system that pursues a goal over several steps, choosing its own steps and using tools such as email, tickets, or databases. |
| Human-in-the-loop | A design in which a person approves, reviews, or handles exceptions in an AI process. |
| Hallucination | Output that sounds right but is invented or wrong. |
| Prompt injection | Text in content an AI reads that tells it to do something other than its task. |
| Drift | A change in the inputs or in the AI's behavior over time that degrades results. |
| Shadow mode | Running an AI beside the people, comparing its output with theirs without using it. |
| Automation bias | The tendency to over-trust automated output and check it less. |
| NIST AI RMF | A voluntary US framework for managing AI risk, with the functions Govern, Map, Measure, and Manage. |
| ISO/IEC 42001 | A certifiable management system standard for AI. |
Body of Knowledge Definitions
Workshops and Newsletters on AI
The site already has material on AI in quality and operations. These pages give the wider picture and sit alongside the guides above.
Connected Quality and AI
A workshop on connecting quality data and AI.
From Juran to AI
Rethinking quality for today’s world.
The Quality Immune System
How AI changes the value chain into predictive ecosystems.
Low-Code Innovation and Power Apps
Accelerating quality with low-code tools.
Leadership 4.0 Digital Change
Leading people through digital change.
Data Quality in Complex Environments
Data quality as the base for any AI use.
Beyond Chatbots: AI Agents on the Shop Floor
How protocols make agents work on the shop floor.
When AI Goes from Advisor to Operator
Managing autonomous agents in supply chains.
The AI Hype Cycle and the Illusion of Operational Gains
A sober look at claimed gains.
The Illusion of the Automated Checkpoint
AI needs discipline, not just data.
When AI Logic Fails
Building immunity into your operations.
The AI Oracle Fallacy
Why prediction is not proof.
AI Operations Hub: Frequently Asked Questions
Is this hub about building AI?
No. It is about running AI in operations: choosing where to use it, building the business case, governing the risk, designing the human role, and managing agents. It uses quality and operations methods and does not teach programming or model building.
Where should an organization start with AI in operations?
With the work. Map a process that has a clear pain, remove obvious waste, break it into tasks, and pick a task that is high volume, easy to check, and low in risk if it goes wrong. Set a baseline, name an owner and a reviewer, and run a small pilot with a stop rule.
How does AI relate to Lean and Six Sigma?
AI is a new tool for the same goals. Lean and Six Sigma discipline, such as standardizing, measuring, controlling variation, and mistake-proofing, comes before and alongside it. Many AI controls, such as sampling, FMEA, control charts, and change control, are familiar quality tools.
Is this legal or compliance advice?
No. It is educational material written from a quality and operations perspective. AI rules differ by country, region, and industry and are changing. Check the laws and standards that apply to you, and have qualified people review high-risk uses.
Which frameworks does the hub refer to?
The NIST AI Risk Management Framework, ISO/IEC 42001 and ISO/IEC 23894, the EU AI Act, and the OECD AI Principles, as common references, with human-factors research on automation and quality methods such as FMEA and acceptance sampling.
Sources and Further Reading
- National Institute of Standards and Technology, Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1 (2023), and Generative AI Profile, NIST AI 600-1 (2024).
- ISO/IEC 42001:2023, Artificial intelligence: management system; ISO/IEC 23894:2023, guidance on risk management; ISO/IEC 22989:2022, concepts and terminology (check current editions).
- Regulation (EU) 2024/1689 (the Artificial Intelligence Act), as amended in 2026; check the Official Journal for the current text.
- OECD, Recommendation of the Council on Artificial Intelligence (2019, updated 2024).
- Parasuraman, R., Sheridan, T. B. and Wickens, C. D., “A model for types and levels of human interaction with automation,” IEEE Transactions on Systems, Man, and Cybernetics, Part A 30(3), 2000.
- Bainbridge, L., “Ironies of automation,” Automatica 19(6), 1983.