FRAMEWORK FIELD GUIDE
The Human Moat is a framework used on Frameworks with Jordan West to examine five skills: trust, taste, judgment, steering, and agency. It asks what deserves deliberate practice as AI tools become more capable. It is a thinking tool, not a guarantee that a job or skill is protected from automation.
When a tool can produce more answers, plans, and drafts, the next question is what you should get better at. The Frameworks AI episode explores that problem through Kasparov, Deep Blue, and the Human Moat framework.
This field guide turns the episode's five-part framework into a self-reflection exercise. It is not a validated assessment, a forecast of employment, or a claim that AI cannot contribute to these activities. Its purpose is to help you choose a useful area of practice.
Trust concerns the relationship around the output. Did you keep your commitment? Did you check what needed checking? Will you tell someone when a result is uncertain? A polished answer does not settle those questions.
Try this: choose one AI-assisted deliverable this week. Make the owner, review step, and limitations clear before someone relies on it. Observe whether expectations become easier to manage.
Taste is the ability to make and explain a quality distinction. More available options can make that distinction more useful. It also creates an opportunity to practice comparison deliberately.
Try this: compare three drafts against a specific audience and purpose. Explain why one works better, then remove one element that weakens it. Keep a note of your criteria so you can see whether they improve over time.
Judgment means weighing evidence, consequences, and tradeoffs when an answer is incomplete. A confident output can still omit an assumption that matters to your decision.
Try this: ask what would make a recommendation wrong. Identify one assumption, one downside, and one fact you would want before committing. Use the tool to broaden the examination while retaining responsibility for the decision.
Steering concerns direction. A system can make progress on a task whose value you have not examined. Before optimizing the execution, define whose problem you are solving and what useful change would look like.
Try this: rewrite a task as a desired outcome. Replace 'generate ten reports' with the decision those reports are meant to inform. Then ask whether ten reports are necessary.
Agency is taking responsibility for moving something forward. It includes choosing a next step, testing it, and adjusting when reality disagrees. Generating a plan is only part of that process.
Try this: take one stalled idea and name the smallest reversible action you can complete this week. Specify the owner and a review date. Judge the experiment by what it teaches you as well as what it produces.
Write down one recent example for each skill. You can mark each area as strong, inconsistent, or untested, but use examples rather than a flattering overall score. Choose the area where a small improvement would matter to something you are actually building.
Show descriptions use the official pages below. Comparisons and exercises are Frameworks editorial assessments. Other shows and hosts are not affiliated with or endorsing Frameworks.