Operational AI Readiness, the show. Short episodes, one obstacle each: what goes wrong, what good looks like, and one thing you can do this week. Episode 2 of 12.
How many AI systems does your organization run? Not how many it bought or approved, but how many are actually shaping work and touching data today. In this episode (about 18 minutes), Viktoria Bakos, author of Operational AI Readiness, looks at what happens when the honest answer is “we’re not sure.”
What the episode covers
Two public audits show two ways an inventory fails. In the first, the AI never makes it onto a list. Ontario’s Auditor General found that the provincial public service had one approved generative AI tool, yet from April to August 2025 other popular AI websites made up 94 percent of staff use, and 244 of the roughly 400 AI websites staff visited were rated unsafe or unsecured (Office of the Auditor General of Ontario [OAGO], 2026, pp. 12, 14). The ministry agreed with the recommendations and says it has since blocked high-risk sites; the auditors have not yet verified that (OAGO, 2026, p. 33).
In the second, the list exists but is wrong. The US Government Accountability Office found that 15 of 20 federal agencies’ AI use-case inventories had incomplete and inaccurate data, some missing the responsible office or a point of contact, and two listing things that turned out not to be AI (U.S. Government Accountability Office [GAO], 2023). Agencies must still inventory their AI use cases every year and publish a public version (Office of Management and Budget, 2025, p. 12).
Key idea: an AI inventory
An AI inventory is a short record of every AI-enabled workflow an organization runs: what it does, what it is for and not for, who owns it, what it depends on and what it can affect. It is the first control because everything else, from oversight to audit, assumes you know what exists. For ordinary people it matters because an organization that does not know which AI tools touch your data cannot protect that data or explain what happened to it.
Three key takeaways
- AI arrives through side doors. Vendor features switch on in updates, browser tabs open, teams share prompt templates. None of it feels like “launching an AI system,” so nobody fills in a form.
- List workflows, not products. Risk lives across the whole workflow, from inputs and retrieval to tools and human review, so a product name tells you almost nothing (Bakos, 2026, pp. 23–24). In the book’s words, an organization that cannot list its consequential AI workflows “is not yet governing AI. It is merely surrounded by it” (Bakos, 2026, p. 34).
- A good inventory finds gaps. The readiness assessment warns that writing an inventory record will surface undocumented connectors and prompt variants: “That is the point.”
The one thing to do this week
Write a one-page inventory record for one AI tool your team uses: a plain description of the workflow, its purpose and excluded uses, a named owner, its moving parts (model, prompts, retrieval sources, connectors and tools), the areas it can affect, and how its outputs are reviewed. Ask the people who use it to check it. Then add a final line called “known gaps.” If you fill that line, the record is working.
If AI is used on you: you can ask whether AI tools are used when handling your information, and whether they are on an approved list. US federal agencies publish an AI use-case inventory each year that anyone can read.
Read the related article
For a deeper look at discovering the AI already running in your organization, read How Much AI Is Actually Operating Inside Your Organization?
Go further. Operational AI Readiness is FairByDesign’s free field manual for AI that is already live. Chapter 1, “What Operational AI Readiness Is,” and Chapter 2, “Where Risk Appears in Live AI Systems,” explain why a list of products is not an inventory; the inventory checklist opens Chapter 4. Read the chapters free. The first 10 readers who send feedback can email their own questions to info@fairbydesign.org.
Your next step: take the free readiness assessment on one AI workflow. Its first domain is inventory and scope, and it shows your weakest area, three next moves and the chapter to read. Nothing is stored; it is a self-assessment, not legal advice or an audit.
FairByDesign is reader-funded. If this helped, get the free Mercury Brief or see ways to support. Support never buys editorial control.
References
Bakos, V. (2026). Operational AI readiness: A practitioner guide to governing AI that is already live (Free ed.). FairByDesign. https://fairbydesign.org/operational-ai-readiness/
Office of Management and Budget. (2025, April 3). Accelerating federal use of AI through innovation, governance, and public trust (M-25-21). https://www.whitehouse.gov/wp-content/uploads/2025/02/M-25-21-Accelerating-Federal-Use-of-AI-through-Innovation-Governance-and-Public-Trust.pdf
Office of the Auditor General of Ontario. (2026). Use of artificial intelligence in the Ontario government [Special report]. https://www.auditor.on.ca/en/content/specialreports/specialreports/en26/2026_AI_EN.pdf
U.S. Government Accountability Office. (2023). Artificial intelligence: Agencies have begun implementation but need to complete key requirements (GAO-24-105980). https://www.gao.gov/products/gao-24-105980
Leave a Reply
You must be logged in to post a comment.