Skip to content

IntermediateLesson 5 of 57 min read

Keeping control: human validation

A course from the Academy: Automating without writing code

Go to the quiz ↓

Last lesson, and the most important one. The question isn't "does it work" but "what happens when it gets it wrong".

The three levels

Fully automatic. Reserved for what is reversible and inconsequential: sorting, labelling, archiving, filling in a tracking spreadsheet.

Preparation, then validation. The system prepares a draft, a human reviews and triggers it. That's the right setting for almost anything that leaves the organisation: customer replies, quotes, publications.

Suggestion only. The system proposes, the human does. For anything that carries legal or financial weight.

The question that settles it

"If this step gets it wrong one time in fifty, what does it cost, and would I spot it?" A botched email sort can be fixed. A wrong quote sent to a client cannot.

The drift to watch for

An automation with validation always ends up becoming automatic in practice: after three hundred correct drafts, you validate without reading. That's human, and it's predictable.

If you know it will happen, two honest responses: either you accept the move to automatic and put the safeguards in place to match, or you make validation impossible to rush, for instance by requiring the amount to be typed out rather than clicking "validate".

What doesn't get automated

Responsibility. Whatever the chain, your name is at the bottom of the sent message. No tool takes that part over.

Complete the lesson

A 4-question quiz: 3 correct answers are enough, plus an exercise to practise. It opens in the Academy, on this same lesson.

Take the quiz and complete the lesson Free. With an account, your progress is saved and leads to the course certificate.

The course: Automating without writing code

  1. What deserves to be automated8 min
  2. The principle: a trigger, then steps.8 min
  3. The tools, and which one to start with8 min
  4. What breaks, and how to spot it8 min
  5. Keeping control: human validation7 min

The whole Academy →The AI glossary