Should AI do this task?
Answer seven questions about one task. You get a verdict, the reasons behind it, how much approval it needs, and what to do first.
Your result
Answer all seven questions.
0 of 7 answered.
How it is scored: seven questions, each weighted by how much it matters (risk counts most). Any task where a mistake is irreversible, the rules do not exist, or the work is mostly judgment is marked KEEP HUMAN, whatever the total.
What the three verdicts mean
- GOOD FOR AI. The task repeats, has clear rules, uses digital information, needs little judgment and is cheap to get wrong. Even here we recommend approving every output at first.
- AI WITH APPROVAL. The employee can do most of the preparation, but a person should approve what leaves the company. Most real business tasks land here, and that is fine.
- KEEP HUMAN. The risk is too high, the process is not defined, or the work is mostly judgment. A person stays in charge. AI can still help with research or drafts.
How the score works
Each answer is worth 0 to 3 points, multiplied by a weight. Risk counts three times, rules, data, judgment and exceptions count twice, frequency and tools count once. The total becomes a score out of 100.
The score is not the whole story. If a mistake would be irreversible, if there are no real rules, or if the work is mostly judgment, the verdict is KEEP HUMAN whatever the total, because those three are not made up for by being frequent or digital.
What the score cannot tell you
It cannot see your actual process, your customers or your tolerance for mistakes. It is a structured way to think, not an audit. Two honest people can answer the same questions differently, so ask someone who does the task to answer them too.
Next
If you want a list of candidate tasks first, use the AI Employee Opportunity Map. To see what the work would cost, use the cost calculator. When you are ready, build the free blueprint.