The classroom
Learn it before you buy it.
Ten parts, from beginner to advanced. One question per lesson, a few minutes each, in plain English. Free, and no sign-up.
An owner who understands the thing buys less of it, and buys the right one.
Start learningStart hereWhat can AI actually do for a small business today?Part 1, Start here. 3 minStartNot sure where to start?Six quick questions about your business, then a path of lessons picked for you.Get your AI plan
Ten parts, beginner to advanced.
169 lessons so far. Take them in order, or open any part.
Or look up a term, A to Z
Beginner
Parts 1 to 3. What AI is, where it saves a small business time, and keeping your data safe.
Intermediate
Parts 4 to 7. Buying it and running it, giving it your knowledge, and putting it to work.
Advanced
Parts 8 to 10. Cost, quality, and AI in your industry and on the shop floor.
Earlier lessonsTwelve short units, show themHide them
- Lesson 1What a language model isIt guesses the next word. That is the whole trick.Open
- Lesson 2What an agent isAn agent is a model with hands.Open
- Lesson 3“Trained on your data”Almost nobody trains anything.Open
- Lesson 4Where it gets it wrongIt will be wrong, and it will sound fine.Open
- Lesson 5Prompting, plainlyBrief it like a new hire.Open
- Lesson 6Why it forgetsIts memory is a desk, not a filing cabinet.Open
- Lesson 7Fine-tuning or retrievalTeach it the voice. Hand it the folder.Open
- Lesson 8How it reads a documentA scanned PDF is a photograph.Open
- Lesson 9What happens to what you typeWhere the words go after you press send.Open
- Lesson 10Open or closed modelsThe one you rent, and the one you run.Open
- Lesson 11The human in the loopWhoever signed it still signs it.Open
- Lesson 12The words vendors useThe vocabulary, in one page.Open
- Lesson 13Why it guesses the next wordAutocomplete, taken seriously.Open
- Lesson 14Why it sounds so sureFluent is not the same as right.Open
- Lesson 15What it cannot knowIf it was not on the desk, it is a guess.Open
- Lesson 16Reasoning models, plainlyIt thinks out loud first. That helps, and it costs.Open
- Lesson 17What an agent can touchEvery tool you hand it is a door.Open
- Lesson 18Where an agent stops and asksDecide the stop points before it starts.Open
- Lesson 19Agents that run on a scheduleThe job that runs while you sleep.Open
- Lesson 20When you don't need an agentA rule is cheaper than a robot.Open
- Lesson 21Training, tuning and readingThree words that vendors use as one.Open
- Lesson 22What “learning from you” meansIt is taking notes, not going to school.Open
- Lesson 23Will it leak to someone elseYour file is not in the model. It might be in the logs.Open
- Lesson 24Made-up facts, and where they come fromIt fills gaps the way it fills sentences.Open
- Lesson 25Wrong with numbersIt reads figures as words.Open
- Lesson 26Wrong about dates and todayIt does not know what day it is.Open
- Lesson 27Catching it: cheap checks that workCheck the claim, not the prose.Open
- Lesson 28Give it the job, not a hintBrief it like a new hire, in writing.Open
- Lesson 29Show it one good exampleOne example beats a paragraph of rules.Open
- Lesson 30Tell it what to do when unsureSay what it should do when it doesn't know.Open
- Lesson 31The prompt you write onceStanding instructions, so nobody retypes them.Open
- Lesson 32The context window, plainlyThe desk has edges. Here is where they are.Open
- Lesson 33What memory features really doRemembered means written down somewhere.Open
- Lesson 34The middle of a long documentPage one and the last page get read. The rest gets skimmed.Open
- Lesson 35Retrieval, step by stepFind the page, then ask the question.Open
- Lesson 36What fine-tuning is good forVoice and format, not facts.Open
- Lesson 37Keeping the answers currentChange the document, change the answer.Open
- Lesson 38Scans, photos and handwritingA blurry digit is a guess in a suit.Open
- Lesson 39Tables, forms and layoutIt reads the words and loses the grid.Open
- Lesson 40Spreadsheets and the numbers in themGive it the cells, not a picture of them.Open
- Lesson 41Drawings, photos and what it seesIt describes a picture. It does not measure one.Open
- Lesson 42Who can see itThree sets of eyes, and which ones matter.Open
- Lesson 43Is it used to train the next modelDepends on the plan, not the promise.Open
- Lesson 44Where it is stored, and for how longEvery conversation is a file somewhere.Open
- Lesson 45Settings to change on day oneFive switches, ten minutes.Open
- Lesson 46Running a model on your own machineNothing leaves the building. Something has to run it.Open
- Lesson 47What you give up with eachClosed buys you the best model. Open buys you the keys.Open
- Lesson 48Small models, and when they are enoughMost jobs do not need the biggest brain.Open
- Lesson 49Draft, review, sendThe shape that works, and why it keeps working.Open
- Lesson 50What to let it do aloneReversible, cheap and watched.Open
- Lesson 51Who signs, and what they are signingApproval is a job. Give it to someone.Open
- Lesson 52When the reviewer stops readingThe tenth good draft is when the eleventh slips through.Open
- Lesson 53“Agentic” and “autonomous”Two words that mean: it can act without asking.Open
- Lesson 54“Enterprise-grade” and “secure”Adjectives are not a security review.Open
- Lesson 55“Powered by”, “built on”, “trained on”Which one wrote the answer, and which one wrapped it.Open
- Lesson 56“Hallucination”, “grounding”, “guardrails”Three words for one problem and two half-fixes.Open
Prefer it in one sitting? Every lesson is written out on one page.
Bring the question to the call.
Thirty minutes on your business and where AI might earn its keep in it. We'll tell you what we would look at first, and what we would leave alone. Vendor-agnostic, no strings attached.