The Write-Down Test: Find AI Agent Processes as an Executive
In short
If you can write down a work process step by step, you have identified a concrete candidate for an AI agent. Seth Godin's write-down test becomes an executive tool: it reveals automation potential, secures operational knowledge, and prepares delegation without building your own AI team.
The question sounds almost too simple: Can you write down what your team does all day? If you can describe a workflow step by step, you have identified a concrete candidate for an AI agent. The idea comes from Seth Godin and is often misunderstood as a threat. For executives, it is the opposite: a sober diagnostic that makes automation potential visible without building an internal AI team.
The Core of the Test: Writeability as an Automation Signal
Seth Godin formulated the test on the 'Diary of a CEO' podcast. His argument: If you can precisely write down an activity, you can train, delegate, or automate it. This turns the write-down test into a practical filter for decision-makers. Instead of speculating about abstract AI capabilities, you examine the concrete structure of your workflows.
"You want a job where nobody can write down what you do all day"
For you as a leader, this does not mean replacing people. It means tackling two risks at once. A process that nobody can write down is neither automatable nor protected against the departure of the knowledge holder. This is exactly where AI agents become relevant: they need clear instructions to operate reliably.
Three Questions as an Executive Self-Test
- Question 1: Is the workflow predictable?
- If the steps happen in a known sequence, that is a strong signal for an AI agent.
- Question 2: Are there clear rules for edge cases?
- The fewer exceptions and special cases, the better. Unclear edge cases need human judgment.
- Question 3: Is the required data available?
- Without accessible data, an AI agent can neither learn nor decide. Data availability is a central prerequisite.
If you can answer all three questions with yes, the process is a clear candidate for an AI agent. If you hesitate on any question, you have found a concrete starting point for process improvement. The Swiss AI Podcast puts it bluntly: AI is a management topic, not a pure IT project. Structure before tools.
From Write-Down to Delegation: Securing Operational Knowledge
Writing down processes is not an end in itself. It is the foundation for securing operational knowledge and preparing delegation. Many Swiss SMEs carry critical knowledge in the heads of a few key people. This is not a problem if you treat documentation as a routine. The write-down test turns that risk into a concrete action list.
95%
of processes are not AI-agent-ready according to our readiness check
First step
Start with the process you know best. If you can write it down step by step, you have your first AI agent candidate. Everything else is worth a conversation.
Why Writing Down Is Not a Threat
The write-down test is often misunderstood as a fear test: if you can describe your job, you are replaceable. For a company, that is the wrong perspective. Automation does not replace people; it takes over repetitive knowledge work. Humans remain responsible for judgment, relationships, and exceptions. AI as leverage instead of labor means investing time in structure and gaining capacity for what matters.
Your Next Step in Three Sentences
- Choose one process that regularly costs you time.
- Write down the steps as precisely as you can.
- Check the three questions and mark any gaps.
Once you run the test, you will be surprised how many workflows are already ready for an AI agent. The rest is implementation.
Frequently asked questions
- What is Seth Godin's write-down test?
- The test states that if someone can write down what you do all day, an AI can be trained to do it. For executives, it becomes a diagnostic tool for automation potential.
- Is the test a threat to employees?
- No. It is a sober tool to separate repetitive work from human judgment. The goal is not headcount reduction but capacity for higher-value tasks.
- How do I identify a process suitable for an AI agent?
- Three questions help: Is the workflow predictable? Are there clear rules for edge cases? Is the required data available? If you can answer yes to all three, the process is a candidate.
- What should I do if a process cannot be written down?
- Then you face a double risk: it is neither automatable nor protected against the departure of the knowledge holder. Start documenting the steps, even if incomplete.
- Do I need an internal AI team to deploy AI agents?
- No. You can start without an internal team if you document processes cleanly and work with experienced partners. Clarity about your workflows matters more than technology.
- How does the test relate to operational knowledge?
- Every documented process is secured operational knowledge. It protects you from losing key people and creates the foundation for delegation and automation.
Sources
Chris Jon Graf · AI Strategist & CEO
For over 15 years, Chris Jon Graf has built productive Swiss software platforms and advised executive teams on the strategic deployment of artificial intelligence. Would you like to evaluate this topic for your company?