01What This Guide to Using Claude AI Covers
How to use Claude AI like the top 1% means asking for much bigger outcomes than most people do. In this episode, Julian Goldie breaks down the three moves shared by Felix Reisberg, an engineer and manager at Anthropic who helped build Claude.AI and Claude Code. Felix says most users hand Claude one small job at a time, like writing an email or fixing a line of code.
That misses what the tool can really do.
Felix points out that Claude may have helped find a new planet this month alone. Meanwhile, many people still use it to fix the tone of an email. The gap comes from what you ask for.
This guide shows you the three moves — altitude, depth, and range — that change your results completely.
02The Three Moves: Altitude, Depth, and Range
Felix splits ambition into four types: speed, altitude, depth, and range. Speed is the one everyone already knows — doing the same work faster. He skips it because you already get that part.
The other three are where the real gap is.
- Altitude: Go up a level. Instead of asking for the task, ask for the result you want after the task.
- Depth: Get Claude to work like the best person in the world at that job. Then keep pushing until even that expert might learn something.
- Range: Look at the work through different fields — more fields than you could ever afford to bring in yourself.
03How to Set Up Claude for Bigger Goals
Before you can use these moves, you need to give Claude real access to your data and tools. Felix says if Claude can only see what you paste into a box, you will stay stuck at small tasks. Setting up access matters as much as the prompt itself.
- Give Claude easy access to your data — like crash reports, customer numbers, or website analytics.
- Let Claude use real tools the pros would use — like databases or statistics, not just quick tables.
- Set up outside checks — a tool Claude can run to see if its result is right.
- Make sure Claude can install tools and connect to your systems through connectors or other ways.
04Using Altitude in Your Own Business
Altitude means you ask for the outcome, not just the task. Felix gives his own example: early last year he was asking Claude to fix single crashes in his app one at a time. Later he changed the ask to "keep the app crash-free." That is a huge jump — one job versus an ongoing result.
To use altitude, ask yourself: What do I want to happen after this task is done? For a quarterly review deck, don't just ask Claude to make slides. Ask it to figure out the three most urgent decisions and start working on options for each one. Or ask it to design a better way to achieve the meeting's goal than a deck every three months.
Felix gives a simple test: if you are going to have a meeting about the thing, go up a level.
05Using Depth to Get Expert-Level Work
Depth means you get Claude to work like the best person in the world at that job. Felix has five ways to do this. The three that matter most are: first, make Claude use the real tools the pros use — like real databases instead of a quick table.
Second, make Claude check what you think is actually true. Give it raw data and ask it to test your belief. Third, use an outside judge — get a check from outside the model to see if the result is right.
Felix shares a prompt he uses all the time. Before Claude starts, ask it: What would a top pro in this field do that I'm not asking for? What raw material would they start from? What methods and tools would they use?
What besides you or me would show the right result? Then tell Claude to rewrite your prompt to ask for exactly that.
06Using Range to Bring in Multiple Expert Fields
Range means you look at the work through different fields. Felix says this is where he sees the most magic. Since Claude can run sub-agents at the same time, you can build a whole team for one task.
To find the right fields, use two tricks from a book called The Art of Innovation. First, swap the noun for a verb. A quarterly review isn't a deck — it's getting busy people to believe a story and make a choice.
Second, build the crew. Imagine the deck has to be amazing or aliens blow up the earth. Who would you bring?
For his deck, Felix would bring an analyst, a journalist, and an editor. He would also add a red team — people who poke holes — like someone who looks for reasons a company is weak. Important warning: Don't ask Claude to write like a journalist. That just makes the writing sound weird.
Instead, ask it to do what a journalist would actually do.
07Results and Comparisons from the Episode
Felix backs up his approach with research and real examples. He leans on a psychology concept called self-efficacy — how much you believe you can do a thing. People who believe they can do more, do more.
Two researchers, Edwin Locke and Gary Latham, went through 35 years of goal research and found that people with a clear, hard goal do best compared to people told to "just do your best."
Felix also shares a story about Sebastian Sohr running the first official marathon under two hours. In the same race, the runner who came second also went under two hours. Seeing someone else do it made it possible for others.
Felix says the same thing is happening with AI now. What was possible last month might be normal this month. He also warns that asking Claude to grow your whole business on its own doesn't work for most companies in 2026 — but the higher you can go before it breaks, the more work comes off your plate.
08Tips, Limits, and Mistakes to Avoid with Claude
Felix gives several important warnings. First, don't let range turn into play-acting. Asking Claude to write like a journalist makes the writing sound weird.
Instead, ask it to use the journalist's methods, not their voice. Second, always use an outside judge for depth. As of October 2026, Felix says get a check from outside the model — a tool Claude can run itself to see if the result is right.
Third, big asks may fail sometimes. Felix points to a book called The Inner Game of Tennis. You play worse when you try too hard not to mess up.
Giving yourself room to fail is how you find the ceiling. A failed big ask still teaches you more than a small ask that worked. Finally, don't wait for AI to settle down.
The models jump every few weeks. If you wait, you are waiting while other people learn the habit of asking for more — and that habit carries to every new model.
09Who Should Use This Approach and Why It Matters
This approach is for anyone who uses Claude regularly — business owners, marketers, content creators, and even engineers. Felix's main example is a quarterly review deck, which is about as normal as office work gets. His best tricks are just questions: What do I want after this task? Who would be the best person in this field?
Who would I bring onto my team? You don't need code to ask any of those.
Setting up access to your data takes effort. But once Claude owns the result, you stop feeding it task after task. You set up once, then you get out of the loop.
That is the time back. Julian says the skill is not the tool anymore — it is how you ask. The people who start asking bigger now are the ones who will get the most from every update after this.
