Free course · The AI Officer Circle
Agents, RAG, MCP, hallucination. You hear these words from vendors, from your board and from your own team. Each lesson explains one of them in a few minutes, in plain language, and ends with the questions to ask before you buy, build or approve anything. No code. No technical background needed.
Join free →Free inside The AI Officer Circle on Skool. Twenty lessons live today.
A one-minute tour of the course. Captions are built in.
How every lesson works
Every lesson has the same shape, so you always know where the useful part is.
1 · Watch
The best short explainer we could find on that word, credited inside the lesson. Often we tell you to watch only the first few minutes and skip the technical part.
2 · Look
The idea in pictures: what it is, how it works, and where the risk sits. Made for people who decide, not people who code.
3 · Decide
Three questions to ask before you buy, build or approve anything that involves the word. This is the part you take into the meeting.
Inside lesson 1, AI agents: the three diagrams
And the decision it changes: when a vendor says "our agent does it", ask which pattern it is, which tools it can touch without a human approving, and who sees it when it gets something wrong.
The lessons
Watch times are the part of the video we ask you to watch. Add a few minutes for the diagrams and the questions.
decide · act
A chatbot answers. An agent decides what to do next, and does it.
The decision it changesWhen a vendor says "our agent does it": which pattern is it, and can it act without a human approving?
About 3½ min watch
create · not just predict
Older AI sorted and predicted. Generative AI makes something new: text, images, voice, code.
The decision it changesIs this tool giving us a score or a draft, and who checks the drafts before a customer sees them?
About 2 min watch
predicts · the next word
It writes by predicting the next word, again and again. It doesn't look the answer up.
The decision it changesDoes this task need fluent language, or a verified fact?
About 4 to 8 min watch
confident · wrong
The AI states something false in the same confident tone it uses when it's right.
The decision it changesWhat does one wrong answer cost here, and is the model allowed to say "I don't know"?
About 5 min watch
memory · limit
How much text the AI can keep in mind at once. Go past it and the start falls out.
The decision it changesWhat happens when our document is longer than the window, and who watches the token bill?
About 5 min watch
thinks · before answering
It works through a problem step by step before answering. Slower, pricier, better at multi-step work.
The decision it changesDoes this task have several dependent steps, and is the extra cost worth the better answer?
About 3 min watch
text · image · voice
AI that reads photos, documents, calls and video, not only text.
The decision it changesWhich of our processes start with a photo, a call or a paper form, and do we have consent to record it?
About 5 min watch
standing · instructions
The standing instructions the AI reads before every conversation, before you type a word.
The decision it changesWhat must every AI answer in our company respect, and who writes and owns those instructions?
About 5 min watch
looks up · then answers
The AI looks things up in your own documents first, then answers, ideally showing the source.
The decision it changesAre our documents current and owned by someone, and does every answer show where it came from?
About 6½ min watch
retrain · or reference
Fine-tuning changes how the model behaves. RAG changes what it knows.
The decision it changesAre we changing what it knows or how it behaves, and who pays to redo the fine-tuning when the model is upgraded?
About 4 min watch
meaning · as numbers
Text turned into numbers that capture meaning, so a system can find "similar" content.
The decision it changesWhere are our embeddings stored, in which country, and can we take them with us if we leave?
About 4 min watch
decides · or follows
A workflow follows steps you defined. An agent decides its own steps toward a goal.
The decision it changesCould we write the steps down in advance? If yes, why do we need an agent?
About 7 min watch
one plug · many tools
A common standard that lets any AI assistant connect to your tools the same way. Like USB for AI.
The decision it changesWhich systems may an AI read, which may it change, and who approves a new connection?
About 2 min watch
AI drafts · a person decides
A person reviews or approves the AI's work at the points where a mistake would matter.
The decision it changesWhat does one wrong output cost, and who is the named human with time to check it?
About 3½ min watch
describe · it builds
Building software by describing what you want in plain language and letting AI write the code.
The decision it changesIs this a prototype to learn from, or a product to run, and who maintains it later?
About 2½ min watch
own it · or rent it
Rent a model through a vendor's service, or download one and run it where you choose.
The decision it changesCan this data leave our walls, and what is plan B if our provider doubles the price?
About 4½ min watch
specialist · generalist
Large models know a bit of everything and cost more. Small ones can be excellent at one narrow job.
The decision it changesIs this a varied job or one task repeated thousands of times, and what does a month at full volume cost?
About 5½ min watch
test · before you trust
A test set of real cases with known good answers, run again every time something changes.
The decision it changesWhat are the 30 real cases we'd test any AI on, and who reruns them when the vendor updates?
About 4½ min watch
used · but not approved
AI your people already use for work without the company knowing or approving it.
The decision it changesWhich tools are our people already using, and which data must never go into them?
About 5 min watch
risk tiers · obligations
Europe regulates AI by risk: some uses banned, high-risk uses with strict duties, most needing only transparency.
The decision it changesWhich of our AI uses touch hiring, credit, health or safety, and who keeps the list?
About 5 min watch
chosen by members
The next lessons come from the words members hear at work this month. Bring the one that came up in your last vendor meeting.
The decision it changesWhatever is on your table right now.
Proposed and chosen inside the community
We don't make the explainer videos. Each lesson uses the best short explainer we could find on that word, and the creator is credited inside the lesson. The diagrams and the decision questions are ours.
Why it's built this way
Owners, directors and senior managers who sign off on AI without being technical. You don't need to build any of this. You need to know what to ask.
Most lessons take under ten minutes, start to finish. Take one before a vendor meeting, or one a week with your coffee.
Every lesson links its questions to the Circle's tools, like the 9 Questions and the Vendor Interrogation, so a word turns into a better decision.
Free inside The AI Officer Circle
Concepts sits in the free tier of The AI Officer Circle, next to the 9 Questions and the community feed. Join, open the Classroom, start with AI agents.
Join free →The Circle starts live on Thursday 19 November, 18:00 Madrid (noon New York).