Skip to content

AI Compass / Knowledge cards

AI process audit

splitting a workflow into software, AI and human

Checked Solo business

AI LABS describes a way of working called Forward Deployed Engineering: a developer sits with the customer, watches the real work and only then builds. For small businesses, this can become a process audit, a written inventory of a workflow with proposals and a price.

What it is for

You find the steps where AI really helps and can quantify the benefit before anything is built.

Steps

  1. Choose the workflow with the largest volume, ideally based on the order or support history.
  2. Sit next to the person who does the workflow every day for an hour and write down every step in its real order.
  3. For each step, ask why it exists. Without an answer, it is often an old workaround.
  4. Classify each step: fixed rule → ordinary software; unclear text or a judgement → AI; costly mistake → stays with the human.
  5. Plan for failure cases, especially places where the model is unsure and answers anyway.
  6. Test with real, already solved cases and read every wrong answer individually.
  7. Quantify the value in money and present it as an audit: workflow, AI steps, value per step.

What it can become

In his podcast, Torben Platzer rates tailor-made AI mini tools for small businesses as the best side income in his ranking (grade 2 on the German school scale). His setup: an input field, a button, a result. Before every generation, the tool checks the price, and failed generations are not charged. What is sold is the build and monthly operation. For comparison he cites €60–80/h for freelancers and €80–180/h for agencies. These are his figures.

Limits

Sources

Status

Checked 11 October 2026 · Maturity new (method described, not tried by me)

Want to try this with your team? That’s what the AI Lab is for. See the AI Lab