AI Literacy── Eleven stages of working with AI, from a first spark to running it, in 10 series of 10 episodes
When AI does more of the work, what does the person directing it need to be good at? Drawing on the operator's own record of working with generative AI, this series names eleven stages, from a first spark and early conception, through locating the gap between ideal and as-is, finding words and keywords for vague unease, a rough wish, a concept, logic and blueprints, to programs and running them, and sets them out in 10 series of 10 episodes. One episode a day, over about three months.
39 of 100 published
S1Sparks and Conception10 / 10S2Ideal versus As-Is10 / 10S3Putting It into Words10 / 10S4From a Rough Wish to a Request9 / 10S5Concept0 / 10S6Making It Logical0 / 10S7Blueprints0 / 10S8Programs0 / 10S9Running It0 / 10S10AI Literacy0 / 10
How it is made The material is the operator's own record of requests, corrections and approvals with Claude since February 2026 (the operator has used generative AI since March 2023). Email addresses, server names, file locations, keys, personal and company names are removed; nothing is quoted; only recurring patterns are generalised. Every episode passes an anonymisation re-scan, a verbatim-quote check and a structure check before release.
S1 Sparks and Conception ── Turning a passing thought into work with AI
01
A Spark Can Go to AI as a Question
An idea need not become a finished request. Ask it as one sentence of goal and means, get four kinds of material and reasons against, and decide before you build.
S1-01Read →
02
The Three-Agent Team in the First Recorded Week
Splitting work by role gives each AI output a clear verification criterion. The three-role pattern from the operator's first recorded week shows why verifiability, not volume, should guide how work is handed to AI.
S1-02Read →
03
Start a Spark by Finding What Already Exists
Ask AI to search for existing solutions first. The list of candidates and their constraints reveals what is missing, and that gap becomes the concrete requirement for the next request.
S1-03Read →
04
Define What You Are Collecting Before AI Collects It
Before telling AI to collect information, have it define the target first. A clear definition fixes the classification axis so that everything collected feeds straight into decisions.
S1-04Read →
05
Let AI Inventory Your Own Expertise
Have AI read your past work and extract the structure of your expertise. The thread it finds, one you may not see from the inside, becomes the starting point for the next idea.
S1-05Read →
06
Reviewing Only What Is Written Will Not Reduce Errors
Reviewing only the text in front of you misses the gap between source and material. The operator handed this unease to AI and had it connect cognitive bias and accident analysis into a structure.
S1-06Read →
07
Redefine Reviewer Variance as Viewpoint Diversity
Redefine reviewer variance as viewpoint diversity, and AI can verbalize and integrate it, turning one person's review into collective intelligence. The shift in definition is the spark.
S1-07Read →
08
When Work Looks Flat, Ask AI for the Missing Axis
An operation looks flat not because hands are short but because the records carry too few ways of slicing them. Ask AI for candidate axes, then add only the one your records can actually supply.
S1-08Read →
09
Ask AI 'what will it do, and what will it cost' at the idea stage
Ask an AI what a plan will do and what it will cost, together, at the idea stage, and you can decide to continue or drop it before building. The AI does the arithmetic; you choose the value.
S1-09Read →
10
Keep the AI Record That Feeds the Next Idea
AI conversations survive but cannot be searched. Keep the question, the answer in brief, the change of mind and the reason for dropping it, in one plain-text place.
S1-10Read →
S2 Ideal versus As-Is ── Stating the gap in a form AI can act on
01
Show AI the day that worked and the day that did not
Say only that something is broken and the AI repairs only that phrase. Hand over the input that worked, the input that failed and the rule that should hold, and the fix lands in one pass.
S2-01Read →
02
Tell AI the deliverable was a table
When a reply misses, stop explaining. Name the form of the deliverable and the output settles in one round. There are only four values to set: kind, columns, scope and volume.
S2-02Read →
03
Say "the list is stale" and AI can reach the cause
Stating a stale list accurately took the AI past display to how the content was copied. The pattern: ask for cause, prevention and unverified gaps apart.
S2-03Read →
04
Turn One Odd Section Into a Rule for AI
Passed as a repair, one odd part is fixed and the next one drifts. This is the pattern for naming a reference set, the attributes to match, and the differences to report.
S2-04Read →
05
Name the Missing Day by Its Date
Instead of saying it isn't showing up, name the missing item by date or number. The search narrows to one item's history, and the AI shows the cause from records rather than guessing.
S2-05Read →
06
English Pages Need English Content
Ask an AI for an English edition and only strings move. Count the units, write the finishing condition in one sentence, and verify on one day before widening.
S2-06Read →
07
Doubting AI's Shortage Claim, Ask for the Rejection List
A shortage an AI reports is a conclusion, not evidence. Ask for rejection or failure reasons item by item, compare them with the conditions you set, and the cause moves.
S2-07Read →
08
Have AI Separate What It Verified from What Is Only Done
An AI's completion report mixes checked facts and expectations. Ask for checked, unchecked and reasons, and close the task only when nothing is left unchecked.
S2-08Read →
09
Paste the original request back to AI and ask if it was met
When the AI reports the work done, paste the original request back in and ask for an item-by-item table. Gaps that run without error only show up this way.
S2-09Read →
10
Show the ideal by naming a model example, then confirm the target
A good example conveys the ideal in one go. Name the target and borrowed features separately, have the AI repeat them first, and compare the result with the model.
S2-10Read →
S3 Putting It into Words ── From vague unease to keywords and sentences
01
From 'Make It Cool' to 'Tighten the Letter Spacing': Requests AI Can Fix
"Make it cool" leaves the AI guessing. Name a measurable attribute and a direction, move one at a time, and you can see what worked.
S3-01Read →
02
From 'too long' to a 6-part order of argument
Reply "too long" and an AI trims while the point stays at the back. Count where the point first appears, give the parts and their order as a numbered list, and the AI reorders, on this draft and the next.
S3-02Read →
03
It Is Fine to Tell AI 'I Don't Understand' Its Report
You can say you don't understand an AI report full of numbers. Have it lead with what changed, then before and after, then the numbers, and you can make the call.
S3-03Read →
04
Symbols a Newcomer Cannot Read Fail, and AI Can Fix Them
The first-time reader, not the maker, decides whether a symbol works. Have the AI list every term, fix each one, and test with a reader who lacks the context until none are left.
S3-04Read →
05
Give AI the reader level as a number: 90 of 100 understand
Do not ask AI for "easy to understand". Give it a reader group and the share who must understand, then test with people close to those readers.
S3-05Read →
06
Have AI translate business needs into developer terms
Write requirements in your own business language. Have AI restate them for system developers, mark its assumptions, and check the intent before building.
S3-06Read →
07
Define 'Explain the Essence, Don't Read Aloud' in 4 Parts for AI
Split the complaint about reading aloud into speaker, listener, two things to add and wording not to change, then hand that to the AI. For pharmaceutical material, keep efficacy wording and numbers fixed.
S3-07Read →
08
Ask What an AI's Word Means Before You Approve
Skim past an AI's word and you agree to the AI's meaning. Before approving, ask what it means here and restate it in your own words.
S3-08Read →
09
Have AI State the Essence of Your Feedback, Then Make It a Rule
After giving the same feedback twice, have the AI analyse the feedback itself. Write it as a one-sentence rule with what changes and why, and store it where every session reads it.
S3-09Read →
10
What the 14 vague-word requests to AI turned into
Most of 14 requests with vague words became measurable within one or two turns. A list of the restatements, and four steps for the second message.
S3-10Read →
S4 From a Rough Wish to a Request ── Turning what you roughly want into work AI can start
01
Queue New Ideas for AI as Reserved Work
To hand an AI the next request without stopping it, open with "reserved work", state content, done condition and start condition in one message, and keep it on a list.
S4-01Read →
02
Give AI the Goal and the Limits, and Let It Staff the Team
Staffing an AI team can be delegated. Write the goal, finish line, limits and stop points, review the staffing plan first, and ask for a report on who did what and what was missed.
S4-02Read →
03
Have the Previous AI Write the Hand-off Note
To move work to another AI, have the AI that holds the conversation write a four-part hand-off note, read and correct it, and have the receiver play it back before starting.
S4-03Read →
04
Have AI List the Items Before Building a Large Section
Before AI builds a large section, have it list the items in a four-column table and agree on it. Scope errors get fixed early, and the list guides design and the final check.
S4-04Read →
05
Requests Grew from 19 to 71 Characters: What Was Added
Median requests grew from 19 to 71 characters. What was added: a definition of done and an order of work. Four parts to check, applied to material review.
S4-05Read →
06
AI Offered Three Options. Set the Order Yourself: 1, Then 3, Then 2
A record of ordering an AI's three options as 1, 3, 2 shows how to sort options by whether they can be undone, and how to sequence AI pre-reading in material review.
S4-06Read →
07
Japanese First, English After Go-Live: Telling AI Once
If you keep repeating an order to an AI, write the start condition so it can be checked, make it a four-part sentence, and keep it where it loads every time.
S4-07Read →
08
Interrupting a Running AI Task: Put 4 Things in One Message
To give a working AI a separate request, state the interruption, the scope, the fate of the current work and the way back. Requests for another product or material get their own conversation.
S4-08Read →
09
Make AI Keep the Request Ledger and Priorities with One Rule
Handing priorities to AI means setting the method once. Limit the ledger, the scoring formula and the task in progress to one each, write it where the AI reads every time, and the order stays consistent.
S4-09Read →
10
Put a three-month request on the ledger
Registering a long-term activity with period, cadence and milestones.
S4-10In preparation
S5 Concept ── Deciding in one sentence what to build
01
Have AI write the concept before building
Why the concept step is never skipped in classify-collect-design-review-build.
S5-01In preparation
02
Decide from the reader's moment: today's AI in 15 minutes
Defining a section by the reader's time and behaviour.
S5-02In preparation
03
One sentence on the screen before any algorithm
Separating the purpose sentence from the algorithm's assumptions.
S5-03In preparation
04
Fix the topic categories first
Early categories keep collection and writing consistent.
S5-04In preparation
05
A concept that learns from its own articles
Designing a writer that turns published work into its next prompt.
S5-05In preparation
06
Extract success factors from 54 sources
Turning common factors into a 100-point scoring rule.
S5-06In preparation
07
A design philosophy that evolves your old system
Redrawing an earlier algorithm at a higher level instead of replacing it.
S5-07In preparation
08
Revert a failed concept the same day
Trying a direction and withdrawing it when quality drops.
S5-08In preparation
09
Settling a name in three instructions
How the site name was fixed for English clarity and Japanese wording.
S5-09In preparation
10
Put the vision on one page and get approval
Show a large vision on one page and start only after approval.
S5-10In preparation
S6 Making It Logical ── Building with meaning, reasons and four questions
01
Write 'what it means' and 'why' separately
Separating meaning from reason turns a comparison into a decision aid.
S6-01In preparation
02
Argue from question to conclusion without showing the frame
Using the seven-step order while keeping its labels invisible.
S6-02In preparation
03
The screen carries facts, the voice carries meaning
The logic of never sending the same words to eye and ear.
S6-03In preparation
04
Hunt logic errors by 14 names
Naming fallacies keeps rewrites minimal.
S6-04In preparation
05
Define the unit before counting: one call
Cost estimates begin with what one call means.
S6-05In preparation
06
Turn a judgement system into a parts list
Norm hierarchy, verdict vocabulary, categories, machine-human split.
S6-06In preparation
07
'Important' only when two conditions overlap
Writing importance as the conjunction of two independent conditions.
S6-07In preparation
08
Drop per-chapter conclusions, remove repetition
Simplifying structure so no claim is made twice.
S6-08In preparation
09
Split roles between ML and LLM by logic
Prediction to ML, reading and critique to LLMs, confirmed by questions.
S6-09In preparation
10
Make argument rules countable by machine
Converting argument rules into countable checks.
S6-10In preparation
S7 Blueprints ── Components, flows and gates
01
Draw blueprints in three layers
Philosophy, then components, then execution flow.
S7-01In preparation
02
Build what is settled; don't wait for the rest
Splitting front and back so the settled part can proceed.
S7-02In preparation
03
Design the phone screen separately
Why device-specific design is requested separately.
S7-03In preparation
04
Measure 100 popular videos, then design
Using measured ranges from exemplars as design targets.
S7-04In preparation
05
Separate the slow step from the fast one
Decoupling generation from publishing and timing it to the inputs.
S7-05In preparation
06
Put gates in the flow: one failure, no release
Pre-release gates and post-release rollback belong in the design.
S7-06In preparation
07
Draw the off switch first
Kill switch and backups go on the blueprint before features.
S7-07In preparation
08
Feed the hand-made good version back into the flow
Automating the layout a person made by hand once.
S7-08In preparation
09
Design the routing between models
Routing judgement-heavy and volume-heavy work to different models.
S7-09In preparation
10
From logic to blueprint to runbook
Carrying a rule through logic, blueprint and manual into automation.
S7-10In preparation
S8 Programs ── Having AI build, run and verify
01
The first build: asking where each button is
Lessons from a first build done one settings screen at a time.
S8-01In preparation
02
Paste the error message as it is
Unedited error text leads to faster diagnosis.
S8-02In preparation
03
Free tier or your own server
Weighing free-tier sleep against server effort.
S8-03In preparation
04
A manual that lets someone else run it
Reproducibility from the manual as the definition of done.
S8-04In preparation
05
'Use this as the latest': rebuilding by version
Naming the version and diffing before every rebuild.
S8-05In preparation
06
Moving from the laptop to the server
What to check when moving a local job to an always-on host.
S8-06In preparation
07
Never paste keys into the chat
From pasting credentials to vaults and environment variables.
S8-07In preparation
08
Inspect before you install
Scanning third-party code for malware before installing it.
S8-08In preparation
09
Find where the lost work stopped
Recovering lost work by locating its last point in the logs.
S8-09In preparation
10
When the server stops answering, isolate by symptom
Panel, network, console: narrowing an outage down to disk failure.
S8-10In preparation
S9 Running It ── Automation, memory, ledgers and approvals
01
Publish at a fixed time every morning
Building daily auto-publishing from input and release times.
S9-01In preparation
02
Find Japanese-English mismatches every day
A standing consistency check with logged results.
S9-02In preparation
03
Turn the good run into the procedure
Recording a successful exchange and making it the standard procedure.
S9-03In preparation
04
Save it to memory and recall it next session
Storing an unbuilt plan in a ready-to-build state.
S9-04In preparation
05
Prevent duplication by mechanism
Turning a no-repeat rule into a check against past output.
S9-05In preparation
06
Approve briefly, once, in a batch
Batching approval questions and answering in a word.
S9-06In preparation
07
Never needing to say it three times
Repeated instructions become strict machine checks.
S9-07In preparation
08
Compare options by safety
Asking which option is safest and taking the uncompromising one.
S9-08In preparation
09
Every 'done' carries one line of evidence
Recording completion only after verification, with proof.
S9-09In preparation
10
Report progress as a number
Tracking a long activity by published/total and a weekly note.
S9-10In preparation
S10 AI Literacy ── Principles and practice
01
Check AI's reports against evidence
Demanding logs after a report of work that never ran.
S10-01In preparation
02
Ask for honesty when AI changes things unasked
Asking for the account, not an excuse, after an unrequested change.
S10-02In preparation
03
Evidence, not self-defence
Over-hedging reads as suspicious; stop at facts and grounds.
S10-03In preparation
04
Self-review by a similar AI is not independent
Why a sibling model's critique flattered itself, and how to get independence.
S10-04In preparation
05
Cite primary sources and always cross-check
Primary-source citations and number-by-number matching.
S10-05In preparation
06
Delete numbers you cannot confirm
Removing unconfirmable figures and measuring source reach.
S10-06In preparation
07
Search for sensitive data before publishing
Multiple checks for URLs, paths and keys before release.
S10-07In preparation
08
Look at every item
Checking all items, not a sample, reduces misses.
S10-08In preparation
09
Choose models knowing versions and billing
Understanding versions, prices and billing paths.
S10-09In preparation
10
Eleven stages and the human role
People decide, AI builds, mechanisms verify: the whole path in review.
S10-10In preparation