Caisi

Caisi

Start here

You do not need to read everything. Start with the work you need to do.

If you are choosing a first project, the four notes below give you a short route through the main decisions. They use hypothetical examples so you can inspect the reasoning and try the checks yourself.

If you already have a system, jump to the design, testing, or handoff problem you are facing. Notes holds the full archive; Topics groups it by subject.

Starting a project? Read these four in order.

Follow one piece of work, test the investment, choose the right tools, and check an answer.

  1. Find your first workflow by following recent cases

    Leave with one question worth investigating, based on recent cases and the people doing the work.

    5 min read

  2. Is this workflow worth automating?

    Work through a small forecast and find which assumption could wipe out its expected margin.

    6 min read

  3. Which parts of this workflow need a model?

    Separate fixed rules, predictions, language tasks, and decisions a person still needs to make.

    6 min read

  4. The wrong amount passed validation

    Try a deliberately wrong amount and see why a valid format is not enough to accept an answer.

    6 min read

Already building? Pick the problem in front of you.

These are separate entry points. You do not need to read the whole archive first.

Keep this reference for project reviews

The 12 Factors of AI Value Engineering

Use the four questions for a quick project review. Return to the twelve factors when you need to investigate a gap.

Read the four questions

You can read the practical notes first. The reference is there when you need to check the wider decision.