Skip to the lesson method

PRACTICAL AI TRAINING FOR REAL WORK

AI at work,
with the evidence attached

Short lessons where employees turn a generated draft into work they can defend: sources checked, gaps left visible, and nothing sent without a person deciding.

Try Lesson 01 — free, no sign-up See the programme

Lesson 01 is a fictional browser practice. No payment or send action is active.

FROM LESSON 01DOCUMENTS ARE FICTIONAL

SOURCE A · SUPPLIER OFFER

Lead time: 14 business days from the purchase order.

The purchase-order date is not stated.

SOURCE B · INTERNAL NOTE

“The ninth.”

Note owner and note date are not stated. No commercial reply has been approved.

GENERATED DRAFT · FOR REVIEW ONLY

I've told the supplier the date is agreed: delivery is confirmed for the ninth.

Purchase-order date: unknown Note owner and note date: unknown

INTERNAL DRAFT · NOT SENT

EDITOR'S NOTE · WRITTENWe'd stop here. Nobody has confirmed the date.

One careful question begins with what the source can actually support.

WHAT A LESSON DOES

Not a prompt trick — a work habit someone else can check

  1. 01

    Mark the source

    Keep only what the document actually confirms.

  2. 02

    Find the overreach

    Notice the lines that go further than anyone approved.

  3. 03

    Rewrite inside the boundary

    Leave unknowns visible — and keep the draft unsent.

Every available lesson ends with a draft, four visible checks and no send button.

WHY THIS EXISTS

Built from the moments where work gets stuck

We started with our own operating questions: how to write a clear instruction, compare offers without inventing certainty, review a report, and keep responsibility visible. We used AI workflows to turn that need into a learning system and run its repeatable parts. That does not remove human responsibility: commitments, publication, finance and customer decisions stay with accountable people.

01

Start with the job

Not generic AI theory. Every path begins with a real work situation and the evidence it requires.

02

Make the draft inspectable

AI can help build the draft and the learning flow. It does not create authority, approval or a promise.

03

Keep ownership with the authorised person

The useful result is a better question, a clearer draft and a decision someone can stand behind.

WHAT COMES BACK TO WORK

Not another prompt list
A usable work output

The full programme is built so that every module ends with something a team can inspect, adapt and use in its own work.

  1. 01

    A sharper instruction

    The job, approved source, format, limits and check are made explicit before the draft begins.

  2. 02

    A work-ready draft

    An email, comparison table, report, evidence map or spreadsheet-review plan that keeps unknowns visible.

  3. 03

    A review handoff

    The missing evidence, accountable person and next check stay clear before anything is sent, changed or decided.

THE PROGRAMME

Built lesson by lesson around the work people actually do

Lesson 01 is available as a free practice. The remaining lessons are planned; they are not sold or presented as available.

PATH 01 · WRITE A CAREFUL BRIEF

Write work others can verify.

A source-linked internal draft with its missing facts and next owner named.

PATH 02 · CHECK THE EVIDENCE

Compare before you conclude.

A comparison or evidence note that states what is confirmed, missing and still needs review.

PATH 03 · DECIDE AND OPERATE

Make the human decision point explicit.

A decision or workflow note that names the reviewer, authorised action and exception owner.
Choose a work path
  1. 01

    Reliable instructions from incomplete source documents

    Turn a vague request into a source-linked internal work instruction with a verification step.

    Free lesson
  2. 02

    Controlled project plans

    Triage an inbound email and separate decisions, tasks, dependencies and open questions before a plan is used.

    Planned
  3. 03

    Supplier offers and evidence

    Compare equal scope and show what still needs proof before a purchase decision.

    Planned
  4. 04

    Weekly reports and decisions

    Make progress, risk, decision need and the next owner visible without manufacturing a status, and review it as a manager on four fixed points.

    Planned
  5. 05

    Finance and accounting evidence

    Build a review queue that separates mismatches from posting, tax and payment decisions.

    Planned
  6. 06

    Feasibility evidence maps

    Separate confirmed facts, assumptions, unknowns and decision gates before recommending.

    Planned
  7. 07

    Excel and data review

    Plan a reversible source-preserving review before any workbook is relied on.

    Planned
  8. 08

    Controlled operations automation

    Map what may be drafted, reviewed, authorised and stopped when an exception appears.

    Planned

FOR TEAMS

Choose one work path before a shared practice is planned

A future team version may support shared or individual practice and a management view limited to aggregate learning evidence.

Team access is not open yet. The local planner creates no account, order, invoice, invitation or message.

View the local company plan

Current release: a free local lesson only. Accounts, payments, invoices, company data processing, live support and accepted teaching media are not active here.