"Everyone is too conservative with AI!" said Claude Code's person in charge: "My approach is more ambitious."

Author: QbitAI

Set some unreasonable goals, then tell the model: This is very reasonable!

This is a small AI usage tip shared by Felix Rieseberg, the head of Claude Code desktop version.

How to do it specifically?

Suppose you want to reduce the website loading time from 10 seconds to 5 seconds, he suggests: Set the goal directly to 0.2 seconds.

The model says it can't do it? Then keep telling it: This problem can definitely be solved, try again, try again.

In Felix's view, we can be bolder when assigning tasks to AI.

He also shared a small tip: Change the nouns in the task to verbs.

For example, the original task is "Create a report PPT," it can be rephrased as:

Within 30 minutes, convince a group of busy people of your explanation of the cause and effect of the situation, and make decisions while the situation is still unclear.

In addition to deliberately setting aggressive goals, he also provided several specific practices:

• Don't just let AI "think like an expert." Follow up by asking: What original materials would a real expert look at, what tools would they use, what methods would they adopt, and then write these requirements into the task.

• Don't take your own judgment as the answer directly. Hand it over to AI for verification, let AI go back to the original materials, look for supporting evidence, and possible counterexamples.

• Don't let AI score itself. Code runs in practice, predict results with new data, and ultimately let real results speak.

Here is a detailed tutorial on how Felix uses Claude, polished.

Enjoy～

Using AI, we can be more ambitious

I work on the Claude team.

What’s interesting about this job is that almost every day I learn that something has been accomplished with the help of Claude—sometimes even independently by Claude.

Some things I never thought of, some I originally thought were impossible; and some, I just wasn't curious enough or didn't take the time to try.

Just this month, Claude may have assisted in discovering a new planet; I saw the first AI music video that I truly liked; we even started discussing whether a new era of mathematics is coming.

The model's various capabilities do not advance uniformly. As Terence Tao said, they are "amazing in some directions, but still lacking in others."

Some people have already learned to harness the model's amazing capabilities, as if they are truly flying. Watching them reach such heights with the product I created is truly fascinating.

More people are still exploring how to make good use of these newly grown wings. They are still pondering: Besides answering "What is the capital of France?" or "Can you make the tone of this email a bit gentler?", what else can AI do?

Many AI practitioners and I often keep a phrase in mind: Using AI should be more ambitious.

This phrase has almost become our action slogan. Whenever we see amazing results, we share them with a comment: "We are not ambitious enough."

I also know that while it’s easy to say, it’s hard to actually do. So, here’s a tutorial on "how to be more ambitious."

I categorize ambition into four types: Speed, Height, Depth, and Breadth.

**△**AI Generated

Speed: Completing the same work in less time. This is the most basic way AI creates economic value.

For example, when writing a piece of code, with Claude's help, I can complete it faster. This usage is already quite mature and doesn’t require much thought for more complex methods, so I’ll skip it for now.

Height: Raising the level of abstraction of tasks, focusing on the final results you want to achieve, rather than on individual specific tasks.

Depth: Thinking about tasks like top experts, delving deeper into the underlying data, tools, and issues.

Breadth: Seeking problem-solving ideas from multiple fields, utilizing knowledge, tools, and methods that you normally wouldn’t encounter.

Although I mainly use Claude, these suggestions are not specific to any particular model.

Height: Let the model work at a higher level of abstraction

We often casually refer to this practice as "letting the model work at a higher level of abstraction." When you assign a task to the model, it is actually executing your optimal conception of how to achieve the expected result.

If you want to think from a higher dimension, you might ask yourself—what do you want to happen next? What is the goal behind the goal?

At the beginning of last year, I had Claude help me fix a single program crash issue, and later, I directly asked it to ensure our application always runs without crashing.

Let’s look at a routine office task * (if I may say, it’s even a bit dull) *.

Suppose you are at Acme Company and need to prepare a PPT for the quarterly business review (QBR). Give Claude a bunch of data, and it will be happy to help you create it; maybe you already have a template, and it just needs to fill in the content.

This is really incomparable to discovering a new planet. But for the same quarterly report, what else can we make AI do? We can start from the most basic requirements and gradually raise the ambition:

**△**AI Generated

Step 1: Let it understand what happened in the business.

Most reviews need to answer these questions: What changes occurred this quarter? Why? What should be done next?

If you just want to update the numbers in the template, please resist that impulse—hand over as much data as you can to the model, let it truly understand your business, and then ask it:

Tell me what changes occurred this quarter, why they happened, and what actions we should take.

Step 2: Let it identify the decisions that need to be made.

A meeting—at least a good meeting—ideally should conclude with certain decisions. You can ask Claude like this:

Based on all the information you have, determine what the three most urgent decisions are right now, and analyze possible directions for action, listing the reasons for and against each.

Step 3: Let it redesign this working mechanism.

In the past, to conduct a quarterly business review, humans had to carefully sift through data, understand the meaning of the data, and then determine whether the business needed adjustments.

Now, you can try asking Claude:

The purpose of the review is to allow management to identify problems early and allocate resources and personnel appropriately.

Please design a better implementation method than the quarterly PPT, using our own historical data to prove what problems could have been identified earlier, and then let it run in parallel with the existing method for a quarter.

When you are genuinely pursuing a final result, raising the level of the goal is the most effective method; however, its applicability is relatively limited for creative tasks.

A good judgment signal is: If what you are doing requires a meeting to discuss afterward, then it’s worth raising the goal one more level.

Of course, there are still limitations—at least in 2026, for most companies, simply saying "Claude, go make money" doesn’t seem to make things run automatically.

However, my overall observation is: As capabilities increase, the model seems to be able to work at higher levels of abstraction.

Raising the level of the goal can also help people find workflows that can run sustainably without continuous human intervention, so you don’t have to assign new tasks to the model one by one.

To do this well, it often requires changing existing workflows.

For example, to ensure Claude keeps our application from crashing, I first need to make it easy to access crash reports, understand which application version corresponds to which Git commit, and grant it access to Windows, macOS, and Linux computers. This way, it can truly reproduce the problem independently and verify whether the fix is effective.

Depth: Use top expert methods to thoroughly address the problem

Sometimes, you just want a task to be done exceptionally well.

Being more ambitious in "depth" means pushing the model to use the tools, techniques, media, and knowledge employed by the best professionals in the field, and continuously ramping up—until even these experts might learn something new from your results.

Continuing with Acme Company's quarterly business review. Suppose this company is facing the issue of customer churn, meaning customers are continuously leaving over time.

I found that combining the following methods is very helpful:

**△**AI Generated

Persist in using tools employed by top industry professionals

When you ask the model to "work like a professional," it often just swaps in a set of professional vocabulary, while the actual methods remain unchanged.

To truly delve into the issues, you need to use the tools employed by top data scientists and statisticians. When Claude reaches for a Markdown table, insist that it switch to databases and statistical models.

You don't need to know in advance what experts would use—Claude usually knows.

Set "unreasonable" goals, then tell the model they are very reasonable

Do you want to reduce the website loading time from 10 seconds to 5 seconds? Ask for 0.2 seconds directly.

Does the model think something is impossible? Emphasize to it that this problem can be completely solved, and it needs to keep trying.

**△**AI Generated

Challenge the conclusions you have already established

Especially in the business field, many things we believe to be facts are merely rough approximations of reality.

Let the model access raw information and data, and examine those assumptions with a questioning attitude.

For example:

We believe that user churn is an issue in the onboarding process for new users. Please try to reproduce this conclusion from the raw data.

Discover new knowledge

This step easily follows "challenging existing perceptions"—since Claude has already delved into the data, continue to ask it to look for blind spots:

Build a customer simulation model based on our logs to replicate the past year. Then, identify the earliest signals of user churn that we are not yet aware of and validate the predictive capability with data it has not seen before.

Introduce external validation

Do not let the model evaluate itself.

I think that as model intelligence continues to improve, this issue will eventually be resolved. But in October 2026, I strongly recommend that you persist in using external validation.

Ideally, the validation method should be a tool that the model can run on its own and receive immediate feedback.

This simple prompt often helps me:

I want to delve deeper into this task rather than expand its scope. Before rewriting the task, please complete four things:

Describe what top professionals in this field would do that I have not yet requested.

List the most original materials they would start with, rather than directly using content that has already been organized and digested for me.

List the specific methods and tools they would employ.

Explain what else, besides you and me, could validate that the results are correct.

Then, rewrite my prompt to accurately include these requirements.

**△**AI Generated

Breadth: Utilize methods from other fields

Being more ambitious in "breadth" means examining your work from perspectives of different fields.

The fields involved can be more than in the past, as it was often not cost-effective to do so previously.

This idea is not new. Almost every movie about assembling a team of experts spends the first 30 minutes gathering a diverse group of individuals, each with their own strengths.

As AI products make multi-agent systems increasingly easier to operate—whether through multiple threads, coordinators, or directly using sub-agents—I find that introducing ideas from other fields often yields interesting results.

Putting this into practice is not easy; it requires you to be creative.

Before seeing my colleague Alex Albert use Blender to render interior design, I never thought of using Claude to create a 3D environment for my homepage.

To be clear, I used to work in video games!

However, this is precisely where I often see surprises.

How to do this specifically? I would use the following methods.

Try replacing nouns with verbs

This is an old method mentioned by Tom Kelley and Jonathan Littman in "The Art of Innovation."

The purpose of a quarterly review goes far beyond producing "a PPT." What you need to do is:

Convince a group of busy people of a set of explanations about the causes and effects of events within 30 minutes, and make decisions under uncertainty.

Describing tasks with verbs makes it easier for Claude and me to see which areas can provide unique and useful perspectives.

**△**AI Generated

Assemble a cross-disciplinary team

Suppose this PPT must be stunning, or aliens will blow up the Earth. What kind of team would humanity assemble?

To ensure the quality of the final product, I would bring in an analyst, a journalist, and an editor.

If a live presentation is needed, I might also need to find a motion designer and an audio designer.

Introducing a "red team" specifically for error-checking could also be helpful, so I would bring in a skeptic and a forensic accountant.

However, this approach can easily unintentionally turn into role-playing. The result is just a weirder dialogue between you and the model, with no actual improvement in outcomes.

You should use this method to seek out useful fields and techniques. Directly letting the model investigate using the methods of a skeptic researching investment targets is more meaningful than having it write a role like that.

All the methods I introduced earlier require the model to have a computer that can operate, install software, and access various relevant data, whether through MCP connectors or other means.

There are now many tools that can provide these conditions; I recommend you choose one to use.

Self-efficacy: People also need to update their judgments about their own abilities

That concludes my specific suggestions on "how to be more ambitious."

If you are willing to stay a little longer, I would like to discuss why it is necessary for us to address this issue.

Humans have always been limited by their own judgments: what we think is possible and what is not. More troubling is that we often believe we can only excel in very narrow fields.

A famous example is Michelangelo's reluctance to paint the Sistine Chapel. He considered himself a sculptor and lacked identification with the identity of a painter.

△Francisco Goya's "Modo de volar" (A Way of Flying)

[Francisco Goya is considered one of the greatest painters in Spanish history, regarded as a pioneer of Romantic painting and a key figure connecting classical masters with modern art.]

In researching the issue of "ambition," I learned about a related concept in psychology: self-efficacy.

In simple terms, people with higher self-efficacy believe more in their abilities and thus can accomplish more.

There is evidence that self-efficacy can be enhanced through certain methods.

Locke and Latham reviewed 35 years of goal-setting research and reached a clear conclusion: those assigned clear and challenging goals perform better than those merely told to "do their best."

Patrick Collison, the founder of Stripe, maintains a list called "Fast," documenting what humanity has accomplished under bold goals and tight deadlines.

Here are several well-documented methods.

Believe that difficult things can be done

In 1939, George Dantzig was late to a statistics class at Berkeley.

He copied two problems from the blackboard, thinking they were homework, and then submitted his solutions.

In reality, those were two unsolved problems in statistics.

See others accomplish difficult things, then go on to complete difficult tasks

For decades, humanity has been trying to shorten the time to run a mile to under 4 minutes.

In May 1954, Roger Bannister broke this barrier. However, the situation of "only he in the world ran under 4 minutes" lasted only 46 days.

If you pay attention to running, you will find that a similar situation is occurring with the two-hour marathon barrier.

In April of this year, Sabastian Sawe ran the first marathon under two hours in an official event in London. The runner-up in the same race also finished under two hours.

Two-Hour Marathon Barrier: The full marathon is 42.195 kilometers, and humanity's "breaking 2" (running within 2 hours) is considered a landmark physical threshold on par with "four minutes per mile."

**△**AI Generated

Allow Yourself to Fail, Then Challenge Difficulties

"The Inner Game of Tennis" seems to be a business book that everyone loves, and I just discovered it last week.

The core idea of this book is: when you try too hard and overexert yourself, you cannot play your best tennis.

We engineers might still remember when Linux was first announced, Linus described it like this:

I’m doing this for fun, not as grand and professional as GNU—

(Who would have thought that later Linux became a big deal, while the GNU kernel still hasn't been completed...)

Self-efficacy is not a new issue, but it has become extremely important today!

In an era where capabilities leap forward every few weeks, we urgently need some methods to achieve things we have never seriously considered or even believed we could accomplish.