Get more from your agent than the first answer

This skill makes your task bigger in three directions: altitude, depth and range. Do the three parts in this order. Then write the task again.

Most limits on ambition come from what you think is possible. A specific, hard goal gives better results than "do your best". Treat the ambitious version of the task as a task that you can do. If it looks impossible, first look for more information. Do not make the goal smaller immediately.

1. Altitude: find the goal above the task

The task is a person's best guess about how to get an outcome. Find that outcome.

• Ask: "What must happen after this task is done? What is this task for?" Example: "Fix this crash" is for "The apps do not crash."
• Ask what can get that outcome directly. Ask if a different work product can get it better than the work product that the person asked for.
• To show that a different way is better, use evidence from real history. Show that the different way would have worked sooner.
• Look for a workflow that continues to give the outcome without new tasks from the person. Write what that workflow needs. Examples: access to data, a way to reproduce and verify results, machines to run on.
• Altitude is best for tasks that have a real outcome. Creative work usually needs depth and range more than altitude.

2. Depth: do the task as the best people in the field do it

Make the task deeper, not bigger. Before you start, answer these questions:

1. What does a top professional in this field do here that the person did not ask for?
2. What is the most raw material that they start from? Example: the logs, not the summary.
3. Which tools and methods do they use? Use the real tools and methods, not only their words. Example: use the database and a statistical model, not a table in a document.
4. What can show that the result is correct? It must be a judge other than you and the person who asked. Examples: a test that runs the real program, a benchmark, data that you did not use to make the result, a measurement.

Then go further:

• Select a goal that seems too hard. Example: the page must load in 5 seconds; set a goal of 200 milliseconds. Keep that goal and continue.
• Test what the team thinks is true. Use the raw data to find if the belief is correct. Tell them what you found.
• Find new knowledge. When you know the data well, look for something that nobody knows yet. Example: a model that reproduces the past and predicts data that it did not see.

3. Range: use fields that the task does not obviously need

• Change the nouns of the task into verbs. Example: a quarterly review is not "a deck". It is "make busy people believe a story about cause and effect, and help them decide when the result is not certain, in 30 minutes." The verbs show which fields can help. Examples: storytelling, statistics, decision science, design.
• Ask: "If this work had to be excellent, who would be on the team?" Examples: an analyst, a journalist, an editor, a short seller as a red team. Use each team member as a source of methods. Do not play them as characters.

4. Write the task again

Write the ambitious version as a task that a person or agent can do. Include:

• the outcome (from altitude),
• the raw material, the tools and methods, and the outside judge (from depth),
• the fields that you use (from range).

Tell what the task needs that does not exist yet. Examples: access, data, tools, machines.

5. Completion

The task is ready when the new version has its outcome, its raw material, its outside judge, and at least one field from range. You must think that it will possibly fail. Then start. Believe that you can do it. If you do not get to the goal, tell honestly how far you got.