How to Use Claude AI Like the Top 1%

Julian Goldie SEO

Key takeaways

The three moves are altitude (ask for the outcome, not a small task), depth (make Claude use expert methods and checks), and range (bring in methods from multiple fields). They work best when Claude has access to the relevant data and tools; bigger requests can still fail.

Use a clear, ambitious goal instead of “do your best,” and give Claude the access needed to pursue it. For depth, have it refine the request, use raw data and professional tools, and verify results independently; for range, describe the work as actions and combine useful methods from different fields.

Video Flow

  1. 0100:00

    Why small requests limit results

    Small requests such as fixing an email or a single crash leave Claude little room to produce a larger outcome. The guide How to Be More Ambitious is attributed to Anthropic engineering manager Felix Rieseberg, and the opening frames the difference as a matter of what people ask the tool to do.

  2. 0201:51

    Ambitious goals and self-efficacy

    The cited goal research associates clear, difficult goals with better performance than “do your best,” while examples about George Dantzig and Roger Bannister illustrate how expectations about what is possible can constrain effort. The practical example is to ask Claude for an exceptionally fast result, such as reducing a ten-second page load to 200 milliseconds, and encourage it to keep trying.

  3. 0304:06

    Altitude: ask for the outcome

    Altitude means asking Claude to own a continuing result rather than complete one task, such as keeping an app crash-free instead of fixing one crash. That requires access to crash reports and test computers; for business work, the request can rise from making slides to identifying urgent decisions or testing a better way to achieve the meeting’s purpose.

  4. 0408:12

    Depth: work with expert methods

    Depth means asking Claude to work like a top professional by using the tools and raw data that professionals rely on, testing assumptions, and checking results with an independent judge. Before work begins, four questions can prompt Claude to identify missing expert methods, better source material, appropriate tools, and an external way to verify the result, then rewrite the request accordingly.

  5. 0510:56

    Range: combine methods from other fields

    Range means drawing on multiple fields’ methods to improve one task, rather than merely imitating their writing styles. Recasting a quarterly review as helping busy people understand cause and effect and make a decision can reveal relevant roles; a proposed team might include an analyst, journalist, editor, and red-team critic, while a sales script could draw on closing, speech coaching, and research into objections.

  6. 0613:46

    Access makes larger goals workable

    All three approaches depend on Claude having access to a computer, real data, connectors, and the tools needed to act; the crash-prevention example required reports and test machines. A shared workspace with connected tools is offered as one way to provide that access, while the same principle is said to apply to other agents.

  7. 0714:35

    Objections and the value of experimenting

    Technical expertise is not required to ask for outcomes, expert methods, or useful contributors, although ambitious requests sometimes fail and setting up access takes effort. The Inner Game of Tennis is invoked to support allowing room for failure, while the argument against waiting is that the habit of asking for more can carry forward as models change.

  8. 0817:15

    Three practical moves and closing image

    The proposed next steps are to replace a weekly task with its desired outcome, use the four expert-work questions before a demanding task, and describe a major project as an action before asking which fields’ methods could help. Francisco Goya’s painting A Way of Flying serves as the closing image for the claim that people have not yet learned to use AI’s capabilities fully.

Tools & Resources

The AI assistant whose use is the subject of the examples and methods.

  1. 01
    The AI assistant whose use is the subject of the examples and methods.00:00
  2. 02
    The AI assistant whose use is the subject of the examples and methods.02:25
  3. 03
    The AI assistant whose use is the subject of the examples and methods.05:24
  4. 04
    The AI assistant whose use is the subject of the examples and methods.08:15
  5. 05
    The AI assistant whose use is the subject of the examples and methods.11:01
  6. 06
    The AI assistant whose use is the subject of the examples and methods.13:49
  7. 07
    The AI assistant whose use is the subject of the examples and methods.14:45
  8. 08
    The AI assistant whose use is the subject of the examples and methods.17:19

The guide attributed to Felix Rieseberg that supplies the three ambition concepts.

  1. 01
    The guide attributed to Felix Rieseberg that supplies the three ambition concepts.00:33

The Claude app whose engineering is attributed to Felix Rieseberg.

  1. 01
    The Claude app whose engineering is attributed to Felix Rieseberg.00:22

A Claude product named among the apps Felix Rieseberg led engineering for.

  1. 01
    A Claude product named among the apps Felix Rieseberg led engineering for.00:22

A Claude coding product named among the apps Felix Rieseberg led engineering for.

  1. 01
    A Claude coding product named among the apps Felix Rieseberg led engineering for.00:24

A test environment named for checking fixes across operating systems.

  1. 01
    A test environment named for checking fixes across operating systems.05:52

A test environment named for checking fixes across operating systems.

  1. 01
    A test environment named for checking fixes across operating systems.05:54

A test environment named for checking fixes across operating systems.

  1. 01
    A test environment named for checking fixes across operating systems.05:58

A 3D design tool cited as an example of professional software AI can operate.

  1. 01
    A 3D design tool cited as an example of professional software AI can operate.08:55
  2. 02
    A 3D design tool cited as an example of professional software AI can operate.13:25

An untitled prompt framework asking about expert practices, raw materials, methods and tools, and independent verification.

  1. 01
    An untitled prompt framework asking about expert practices, raw materials, methods and tools, and independent verification.10:19

The book cited for the technique of replacing a noun with a verb to identify useful fields.

  1. 01
    The book cited for the technique of replacing a noun with a verb to identify useful fields.11:41

An alternative agent named as compatible in principle with the same access approach.

  1. 01
    An alternative agent named as compatible in principle with the same access approach.14:25

An alternative agent named as compatible in principle with the same access approach.

  1. 01
    An alternative agent named as compatible in principle with the same access approach.14:27

The creator’s shared workspace for connecting agents, memory, and tools.

  1. 01
    The creator’s shared workspace for connecting agents, memory, and tools.14:15

The book cited for the idea that fear of failure can hinder performance.

  1. 01
    The book cited for the idea that fear of failure can hinder performance.15:45

Francisco Goya painting used as the closing image.

  1. 01
    Francisco Goya painting used as the closing image.17:51

Share a little
cookie?

Your cookies help us understand visits and how you use Vigest, so we can make it better. We use Google Analytics. You can use the service without agreeing.

You can change your choice anytime.
Saved in this browser for 180 days.

Privacy notice