Felix Rieseberg, who writes that he works on Claude, published a tutorial on his blog on 13 October 2026 on how to be more ambitious when working with AI models. He sorts that ambition into four types: Speed, Altitude, Depth and Range.
His starting point is that model capabilities do not grow uniformly. Quoting Terence Tao, he writes that models are “astounding in some directions, while inadequate in others”, and that many people are still “struggling to get their new wings to work”.
Business Pill · THE MODEL AND THE HARNESS
A one-minute explainer of the model and the harness: why products built on the same model can behave differently. It teaches the general idea only and says nothing about any person or company in this story.
The key insight: As we read it, the tutorial puts the gap on the user’s side rather than the model’s: Rieseberg writes that capabilities grow unevenly and that many people are still working out what models are good for beyond simple requests.
The Four Types
Speed is “the same work in less time”, which Rieseberg calls the bread and butter of AI’s economic value. He skips it as well established.
Altitude means “going up in abstraction and considering the outcome rather than the individual task.” Depth means “considering a task the way a field-leading expert would.”
Range means “drawing on multiple fields rather than just the one the task might need at the surface”. He writes that he mostly uses Claude but that “my advice is not model-specific.”
Altitude: Ask for the Outcome
To gain altitude, he suggests asking yourself “what do you want to happen next?” His own example: early last year he still asked Claude to fix individual crashes, and eventually he asked it “to simply keep our apps crash-free.”
That took changes to the workflow, he writes. Claude first had to be able to fetch crash reports, map each app version to its git commit, and get access to Windows, macOS and Linux computers so it could reproduce and verify fixes on its own.
His worked example is an imagined company’s quarterly business review. The first step up is to hand the model as much data as possible and ask: “Tell me what changed this quarter, why, and what we should do about it.”
The last step asks the model to “design a better way to do that than a quarterly deck”, show from the company’s own history that it would have caught problems sooner, and run it beside the current review for a quarter.
He notes limits: asking Claude to “go make money” does not seem to work for most businesses by itself in 2026, he writes.
Depth: Use the Experts’ Instruments
For depth, Rieseberg lists five techniques. The first is to insist on the tools the best professionals use: “When Claude reaches for markdown tables, insist on the use of databases and statistical models.”
The second is to pick unreasonable goals and tell the model they are reasonable. If you need a site to load in 5 seconds instead of 10, he suggests asking for 200 milliseconds.
The third is to challenge what you know, giving the model the raw data and asking it to test assumptions, as in his example: “We believe churn is an onboarding problem. Try to reproduce that from the raw data.”
The fourth is to produce new knowledge, for instance by asking for a simulation of customers built from the logs. The fifth is an outside judge: “Do not ask the model to judge itself.” In October 2026, he writes, “I implore you to insist on outside validation.”
He shares the prompt he uses to push for depth:
“I want a deeper version of this task, not a bigger one. Before rewriting it: (1) say what a top professional in this field would do here that I’m not asking for; (2) name the rawest material they would start from instead of what has already been digested for me; (3) name the specific methods and tools they would use; (4) say what, other than you or me, would show the result is right. Then rewrite my prompt to ask for exactly that.”
Range: Bring In Other Fields
Range means looking at the work from more fields than would have been economical in the past, he writes. One technique is to replace nouns with verbs: a quarterly review is not “a deck” but “getting busy people to believe a story about cause and effect, and to decide under uncertainty, in 30 minutes”.
Another is to assemble a crew. For the review he would bring in an analyst, a journalist and an editor, a motion and audio designer if it is presented, and a short seller and a forensic accountant as a red team.
He warns against role play: “Don’t ask the model to write a short seller, ask it to do what a short seller does to research their position.”
All of these methods, he writes, require the model to have a computer that can install software and access to relevant data, through MCP connectors or other means.
Why He Thinks This Is Needed
Rieseberg ties the problem to self-efficacy, the psychology term for how much people believe in their own abilities. He cites Locke and Latham’s review of 35 years of goal-setting research, which found that people given specific hard goals perform better than those told to try their best.
His examples: George Dantzig in 1939 solving two unsolved statistics problems he took for homework; Roger Bannister breaking the four-minute mile in May 1954; and Sabastian Sawe running the first official sub-two-hour marathon in London this April, with the runner-up also under two hours.
He also cites The Inner Game of Tennis and Linux, announced as “just a hobby”.
The Structural Read
As we read it, altitude is a workflow change before it is a prompt: his crash-free example needed crash reports, version-to-commit mapping and test machines before Claude could own the outcome.
As we read it, his depth advice is about checking as much as asking: he tells readers to insist on experts’ instruments and an outside judge rather than letting the model grade its own work.
As we read it, range is where he says he sees the most magic happen, and it is also where he warns against role play that makes transcripts weirder without improving the outcome.
Felix Rieseberg, in his 13 October 2026 post
“We humans have always been held back by what we believe possible”
Three Implications
FOR OPERATORS Rieseberg’s altitude example says the model first needed access to crash reports, git history and Windows, macOS and Linux machines to verify fixes on its own.
FOR ANALYSTS His depth section says to insist on databases and statistical models over markdown tables and to use an outside judge, ideally a tool the model can run.
FOR TEAMS His range section says he has seen interesting results by asking Claude to bring in additional fields as multi-agent systems get easier to run, and that the model needs a computer and data access.
The Business Engineer Lens
This story maps onto the Business Engineer framework The Product Overhang Doctrine.
The framework starts from two curves: “The first is what your underlying technology can do. The second is what your users can actually do with it.”
As we read it, Rieseberg’s tutorial is written for the second curve: he describes some people flying with the models’ capabilities while many others are still struggling to use them. The framework does not discuss his post.
What Is Not Established
The tutorial is a personal blog post. Its Acme Inc example is, in his words, an imagined company, and the post gives no measured results for the techniques.
He writes that the advice is not model-specific, though his examples use Claude.
The Bottom Line
Rieseberg, who writes that he works on Claude, argues that people should aim higher with AI models in three ways beyond speed: altitude, depth and range.
His concrete advice includes asking for the outcome rather than the task, insisting on experts’ tools and an outside judge, and bringing in other fields without role play.
95,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. We read Felix Rieseberg’s 13 October 2026 blog post in full; every technique, example and quotation attributed to him comes from it, including the Terence Tao line and the examples he cites. The Tao link goes to our own earlier piece. We did not use other outlets’ reports.
Sources: Felix Rieseberg: How to be more ambitious: A tutorial for humans with models (13 Oct 2026)
