---
title: "Felix Rieseberg's Guide to Being More Ambitious With Claude"
date: "2026-11-02"
video_id: "7703702868834061591"
tiktok_url: "https://www.tiktok.com/@nathanhodgson.ai/video/7703702868834061591"
creator: "@nathanhodgson.ai"
category: "AI"
tags: ["claude", "anthropic", "felix rieseberg", "prompting", "ai productivity"]
excerpt: "Felix Rieseberg's essay sorts ambitious Claude use into four axes: speed, altitude, depth and range. What each means, his prompt, and what a TikTok got wrong."
---
"How to be more ambitious: A tutorial for humans with models" is a free essay by Felix Rieseberg, an engineering manager at Anthropic, dated 13 October 2026. It argues that people limit AI by what they think is possible, and it sorts more ambitious use of Claude into four axes: speed, altitude, depth and range. A TikTok from @nathanhodgson.ai summarizes it mostly accurately. It garbles his name as "Fall Claude", thins out the meaning of depth, and adds a fraud-investigator example that does not appear in the essay.

## Who is Felix Rieseberg?

The video opens with "Anthropic's engineering lead, Fall Claude, says AI isn't the problem anymore, you are." "Fall Claude" is a speech-to-text error. The first frame shows an X post from Felix Rieseberg (@felixrieseberg) reading "Hi, I'm Felix. I'm an eng lead working on Claude."

His [site](https://felixrieseberg.com) describes him as an engineering manager at Anthropic who has led engineering for Claude.ai, Cowork and Claude Code Desktop. "Engineering lead" fits that. The video's blunt "you are" the problem is the presenter's framing, sharper than anything confirmed in the essay.

## Why does the essay mention Roger Bannister?

The video shows a card reading "3:59.4" and "Roger Bannister · 6 May 1954." The essay uses Bannister to make its main point: he broke the four-minute mile in May 1954 and, according to the essay, was the only person to have done so for 46 days. Seeing a feat done makes it seem possible. The essay's argument is that people should update what they believe models can do.

## What are the four axes of ambition?

### Speed

Speed means doing the same work in less time. The essay names it and skips it because it is well established. The video says "Everyone already does this, so he skips over it in the guide," which is a fair paraphrase.

### Altitude

Altitude means working at a higher level of abstraction and focusing on the outcome instead of a single task. The video says "Give Claude the big goal, not just one small task," and "He doesn't ask it to fix one crash, he asks it to stop his apps crashing at all." That matches the essay, where the author first asked Claude to fix individual crashes and later asked it to keep his apps crash-free.

### Depth

The video says "Aim way higher than normal. If your website takes 10 seconds to load, don't ask for five. Ask for a fifth of a second and tell Claude it can do it." The essay does use this example: go from 10 seconds toward 200 milliseconds rather than 5 seconds, then insist it is achievable. A fifth of a second is 200 ms, so the numbers match.

The video's definition is thinner than the essay's. In the essay, depth means approaching a task the way a top expert would, going a layer deeper into the underlying data, tools and problems. The shared prompt puts it plainly: "I want a deeper version of this task, not a bigger one." A viewer who only hears the video would take depth to mean "demand a bigger number."

### Range

Range means drawing on knowledge, tools and techniques from several fields. The video's example is "checking your business numbers the way a fraud investigator would," with an animation of a duplicated invoice.

That example is not in the essay as far as I could read it. The essay's range examples are:

- a red team made of a short seller and a forensic accountant
- an analyst, a journalist and an editor for a deck
- a motion and audio designer for a presented deck
- Alex Albert using Blender for interior-design renders

A forensic accountant is the closest match. A [besthub.dev write-up](https://www.besthub.dev/articles/anthropic-engineer-s-framework-for-ambitious-ai-usage-altitude-depth-range-54bb5e0625fd) of the essay also lists a forensic accountant and a short-seller analyst, with no fraud investigator. The invoice amounts in the video are invented.

## What is the prompt Rieseberg shares?

The video says "He even shares the exact prompt he uses," but never shows it. In outline, the prompt asks the model, before rewriting your request, to:

- say what a top professional would do that you did not ask for
- name the rawest source material
- name the methods and tools
- say what, besides the model or you, would verify the result

Its key line is "I want a deeper version of this task, not a bigger one." The video's "Launch plan" card, listing "Pro playbook", "Raw user data" and "Outside proof", tracks these elements without explaining them.

## Where to read the guide

The essay is a free post on [Rieseberg's site](https://felixrieseberg.com/how-to-be-more-ambitious-a-tutorial-for-humans-with-models/). The video gives no link, so you have to search for the title. The habit app and "Habitly" website in the video are mock-ups, and the video shows no real Claude session.

## Bottom line

The video gets the structure right: four axes, speed skipped, the crash example for altitude and the 10-seconds-to-200-milliseconds example for depth. Depth in the essay means working like an expert with raw data and outside checks, not just asking for a bigger number. The fraud-investigator example is the video's own, so read the essay and use its prompt directly.

## Sources

- [How to be more ambitious: A tutorial for humans with models, Felix Rieseberg, 13 Oct 2026](https://felixrieseberg.com/how-to-be-more-ambitious-a-tutorial-for-humans-with-models/): the four axes, examples, prompt, Bannister passage and Rieseberg's role.
- [felixrieseberg.com](https://felixrieseberg.com): the essay's title, date and the author's background.
- [Anthropic Engineer's Framework for Ambitious AI Usage, besthub.dev, 15 Oct 2026](https://www.besthub.dev/articles/anthropic-engineer-s-framework-for-ambitious-ai-usage-altitude-depth-range-54bb5e0625fd): secondary write-up whose range examples match the essay's.
- [TikTok post by @nathanhodgson.ai](https://www.tiktok.com/@nathanhodgson.ai/video/7703702868834061591): the video reviewed here.


## Video description and verification notes

**The clip runs 57.98 s (video stream 57.92 s) at 1080x1920 in an MP4 (isom) container, with HEVC Main video at 29.97 fps and about 192 kbps, plus HE-AACv2 stereo audio at 44.1 kHz and about 64 kbps. The file is 1,908,386 bytes at about 263 kbps overall. It was posted at epoch 1793658082, which is 2026-11-02 22:21:22 UTC, by handle @nathanhodgson.ai (nickname "Nathan Hodgson"). The audio is credited as "original sound" by Nathan Hodgson. The description is one line (about 110 characters) repeating the title plus the hashtags #ai #aitools #artificialintelligence #claude #anthropic. The archive capture on 2026-11-03T13:02:47Z showed 29,300 views, 1,555 likes, 18 comments, 133 reposts and 811 saves. I read 9 of the extracted 2-second frames (1, 3, 6, 10, 14, 18, 22, 26 and 29). The transcript has 13 segments and about 165 words. Frame 1 shows a screenshot of an X post from "Felix Rieseberg" (@felixrieseberg, verified badge) reading "Hi, I'm Felix. I'm an eng lead working on Claude. I love watching people take flight and do things they never thought possible," with part of the text highlighted and the burned-in caption "engineering" above the presenter, a man in a black shirt at a microphone against a purple background. Frame 3 shows a phone mock-up of a habit app ("Hi, Alex", "3 of 4", "5 day streak") with a prompt box typing "fix a typo". Frame 6 shows a card reading "3:59.4", a bar running from 3:50 to 4:00, and "Roger Bannister · 6 May 1954". Frame 10 shows a stair graphic with "1 Speed" and a blurred "2" step. Frame 14 shows a "Big goal" tag over a mock website called Habitly ("Build habits that stick."). Frame 18 shows a slider marked "Normal" with its handle at "Way higher". Frame 22 shows the stair graphic with "Speed" struck through and steps for Altitude, Depth and Range. Frame 26 shows three mock invoices, #1041 ($1,250), #1042 ($2,400) and a duplicate #1042 ($2,400) flagged "Duplicate!" under a magnifying glass, with the caption "investigator". Frame 29 shows a Claude-style asterisk icon over a "Launch plan" card listing "Pro playbook", "Raw user data" and "Outside proof". The spoken argument is that Anthropic's engineering lead says the limit is what humans think is possible, and that people should aim higher along four axes: speed, altitude, depth and range. The last transcript segment starts at 52 s and ends at 57.78 s, inside the 57.98 s audio, so nothing is hallucinated past the end. None of the nine frames I read shows a real Claude run, terminal output or the guide's text, and the clip links to no URL.**

## The speaker is Felix Rieseberg, not "Fall Claude"

At 0.0 s the transcript says "Anthropic's engineering lead, Fall Claude, says AI isn't the problem anymore, you are." "Fall Claude" is a Whisper mangling. Frame 1 settles it: the on-screen X post is from "Felix Rieseberg" (@felixrieseberg), and the post says he is "an eng lead working on Claude." The name "Felix" in the post is what the transcript garbled. The sentence "Fall Claude" carries no meaning, and nothing in the video establishes that anyone by that name exists.

The title claim is mostly fair. Rieseberg's own site describes him as an engineering manager at Anthropic who has led engineering for Claude.ai, Cowork and Claude Code Desktop. "Engineering lead" matches his self-description in the post on screen. It does not make him the head of all Anthropic engineering, and the video does not say he is. One secondary summary I found called him "head of Claude Code desktop," which differs from the first-party description, so I rely on his site.

## The guide exists, is free, and has the four axes the video lists

The video's guide is Rieseberg's essay "How to be more ambitious: A tutorial for humans with models," dated 13 October 2026 on his site. I could not reach the X post that secondary coverage links to, and one secondary site (fourweekmba.com) returned a 403 to my fetch, so I cite neither. The essay page is a free blog post. I saw no signup form or paywall in the page content the fetch returned, though that fetch was summarized for me rather than read as raw HTML.

At 11.72 s the video says "His fix is to aim higher with AI in four ways." The essay defines four axes: speed, altitude, depth and range. At 15.32 s the video says "Number one is speed. This just means doing the same job faster," and at 19.1 s "Everyone already does this, so he skips over it in the guide." The essay defines speed as doing the same work in less time and says it skips it because it is well established. The video's "everyone already does this" is a looser paraphrase of "well established," which is a small embellishment and not a factual error.

The video's frame 22 strikes "Speed" through while keeping Altitude, Depth and Range lit. That graphic matches how the essay treats speed: named, then set aside.

The opening line, "AI isn't the problem anymore, you are," at 0.0 s is the presenter's framing. The essay's own framing, as described by the fetched summary and by secondary coverage that ties ambition to self-belief, is that people should update what they think models can do. The blunt "you are the problem" is sharper than any wording I could confirm in the essay.

## The Bannister graphic is accurate and appears in the essay

Frame 6 shows "3:59.4" and "Roger Bannister · 6 May 1954." The essay's Bannister passage, per the fetched summary, says he broke the four-minute mile in May 1954 and that he was the only person to have done so for 46 days, to illustrate that seeing a feat done makes it seem possible. The date and time on the card are consistent with that. I did not independently fetch a Bannister source, so I treat the 3:59.4 figure and the 46-day gap as essay claims, not as verified history. The spoken audio at 5.2 s ("humans have always been held back by what they think is possible") paraphrases the same idea.

## Altitude and depth examples match the essay

At 22.36 s the video says "Number two is altitude. Give Claude the big goal, not just one small task," and at 27.28 s "He doesn't ask it to fix one crash, he asks it to stop his apps crashing at all." The essay's altitude example, per the fetch, is that the author first asked Claude to fix individual crashes and later asked it to keep the apps crash-free. That agrees with the video. Altitude in the essay means working at a higher level of abstraction and focusing on the outcome rather than the individual task. The video's frames 3 ("fix a typo") and 14 ("Big goal") dramatize this with invented mock-ups, a habit app and a "Habitly" site. Those are illustrations, not footage of Rieseberg's work.

At 32.52 s the video says "Number three is depth. Aim way higher than normal. If your website takes 10 seconds to load, don't ask for five. Ask for a fifth of a second and tell Claude it can do it." The essay's load-time example, per the fetch, is to go from 10 seconds toward 200 milliseconds instead of 5 seconds, then insist it is achievable. One fifth of a second is 200 ms, so the arithmetic matches.

However, the video's definition of depth is thinner than the essay's. "Aim way higher than normal" (the slider in frame 18) describes a bigger target. The essay defines depth as approaching a task as a top expert would, going a layer deeper into the underlying data, tools and problems. The shared prompt makes the contrast explicit with the line "I want a deeper version of this task, not a bigger one." The video's frame 29 labels, "Pro playbook," "Raw user data" and "Outside proof," do track the prompt's elements (what top professionals would do, the rawest source material, and an outside check), but the narration never explains them. A viewer who only hears the audio would take depth to mean "demand a bigger number," which is the load-time example and not the whole concept. This is a partial correction to the video's framing, not an error in its facts.

## The fraud-investigator example is not in the essay

At 47.72 s the video says "Number four is range, and this one's fun. Claude borrows tricks from totally different jobs, like checking your business numbers the way a fraud investigator would." Frame 26 animates this with a duplicated invoice #1042 flagged "Duplicate!"

The essay's range definition is drawing on knowledge, tools and techniques from several fields, which matches the first half of the sentence. But the fetched summary of the essay states it does not mention a fraud investigator. The range examples it gives are a red team made of a short seller and a forensic accountant, an analyst, a journalist and an editor for a deck, a motion and audio designer for a presented deck, and Alex Albert using Blender for interior-design renders. A forensic accountant is the nearest relative of a fraud investigator, and a secondary write-up of the essay (besthub.dev) also describes a forensic accountant and a short-seller analyst, not a fraud investigator. The invoice scene, including the duplicate #1042 with $2,400 totals, is the creator's own invention, and the amounts do not come from the essay. I could only read the essay through a summarizing fetch, so "not mentioned" carries that caveat. I recommend treating "fraud investigator" as the creator's paraphrase, close in spirit to the essay's forensic accountant but not its wording.

The video also says range lets Claude "borrow tricks." The essay frames range as the person pulling in other fields' methods, and a related secondary summary says it cautions against role-play that adds noise without better results. The video's phrasing could be read as "have Claude play a fraud investigator," which is the pattern that caution targets. That reading is an inference about the video's looseness, not something the narrator says.

## The "exact prompt" is real, but the video never shows it

At 52 s the video says "He even shares the exact prompt he uses to push Claude even further than most people think is possible." The essay does include a prompt, and the fetched summary quotes its key line: "I want a deeper version of this task, not a bigger one." In outline the prompt asks the model, before rewriting the user's request, to say what a top professional would do that was not asked for, name the rawest source material, name the methods and tools, and say what besides the model or the user would verify the result.

The video never displays the prompt text, so a viewer cannot copy it from the clip. Frame 29's three-line "Launch plan" card is a visual summary, not the prompt. The video also sends viewers nowhere: the description has no URL and the spoken audio gives no site name. A viewer has to find the essay by searching for its title or Rieseberg's name. The `links/` folder for this clip was not available to me, so I did not check any description links.

## Key Takeaways

- **Correction:** The transcript at 0.0 s says "Fall Claude." The on-screen X post in frame 1 shows the speaker is Felix Rieseberg (@felixrieseberg), who describes himself as an eng lead working on Claude.
- **Correction:** The video's range example, "checking your business numbers the way a fraud investigator would" (47.72 s), is not in the essay as far as I could read it. The essay's nearest examples are a forensic accountant and a short seller. The invoice animation is invented.
- **Partial correction:** The video presents depth as "aim way higher than normal" (32.52 s). The essay defines depth as working the way a top expert would, with raw data, professional tools and an outside check, and its prompt asks for a deeper task, not a bigger one.
- **Verified:** The four axes (speed, altitude, depth, range), the decision to skip speed, the crash example for altitude and the 10 seconds to 200 ms example for depth all match the essay on Rieseberg's site, dated 13 October 2026.
- **Clarification:** The "free guide" is a public blog post. The video gives no link, shows no prompt text, and none of the nine frames I read shows a real Claude run.
- **Context:** The clip was posted 2026-11-02 22:21:22 UTC, about three weeks after the essay. At capture it had 29,300 views, 1,555 likes, 811 saves and 133 reposts.

## Resources

- [How to be more ambitious: A tutorial for humans with models, Felix Rieseberg, 13 Oct 2026](https://felixrieseberg.com/how-to-be-more-ambitious-a-tutorial-for-humans-with-models/) establishes the four axes, the crash and load-time examples, the range examples, the prompt, the Bannister passage and Rieseberg's job title. I read it through a summarizing fetch, not as raw text.
- [felixrieseberg.com home page](https://felixrieseberg.com) establishes the essay's title, date and URL.
- [Anthropic Engineer's Framework for Ambitious AI Usage: Altitude, Depth, Range, besthub.dev, 15 Oct 2026](https://www.besthub.dev/articles/anthropic-engineer-s-framework-for-ambitious-ai-usage-altitude-depth-range-54bb5e0625fd) is a secondary republication I cite only to show that its range example is a forensic accountant and a short-seller analyst, with no fraud investigator.
