Dan Shipper plays with AI for a living. That’s his job – so when he gets a bad result, he just tries it again. But for the rest of us, one bad experience creates a dilemma: Do we try again, or give up because we don’t want to waste valuable time?

Dan argues that people get “capability blindness.” We try something once and never try it again if we don’t get a good result. But with the rate of change in AI – if you turn over the same rock a couple months later, you might find something new.

To give you some rock-turning inspiration, we asked him the ways he personally gets high value from AI.

Use case #1: Get psychological feedback

Because this generation of AI tools has a good theoretical mind, it can help you look at something from endless perspectives. Dan uses Claude to enhance his practice of journaling, in order to clock patterns in his behavior and psychology that he should be aware of.

How Dan does it:

1. Feed AI your journal entries. You can use one entry if you want to only untangle your recent events – or you can feed AI  journal entries from a longer period of time if you want to see high-level trends.

2. Prompt AI on whatever you want to know. Maybe you want to understand if there are patterns of behavior you should be aware of, or if you seem to be struggling with the same things on repeat. Ask AI to frankly identify those things and give you tips for improvement.

How to take it a step further:

• Start a Project in Claude for your reflections. Claude’s new Project feature pulls from a knowledge base that you curate with your own reference files. By regularly adding new journal entries and setting intentions for the Project, Claude can give you better and more personalized insights.
  

• Ask Claude to set goals for you. Use Claude to analyze your entries and reflect on how you’ve changed and advanced over a period of time. Then ask it to set you some ambitious goals for the future and lean on Claude as an accountability partner.

Use case #2: Get a ‘referee’ on a contentious issue

Because AI has no skin in the game, it’s great at helping you process tough and emotionally charged situations. Whether it’s a relationship issue, a rough conversation with your boss, or a seemingly lose-lose situation, with detailed prompting it can often give you a good neutral perspective.

How Dan does it:

1. Give AI the backstory. Write a reflection of what you’re dealing with. Provide as much detail as you can on the historical context and the series of events leading up to the issue at hand. (You can use the ‘voice’ feature in ChatGPT to do this without having to write, as well).

2. Ask it what it thinks is going on. You can prompt things like: “Take on the perspective of the other person, and explain how they might be seeing the situation” or “Suggest a few things that you think might have led to this problem.”

How to take it a step further:

• Scrub your prompt of biases. AI is going to mirror your reasoning, so you need to practice some self-awareness. Tighten up the objectiveness. Are you trying to paint the other person in a particular light? Are you including assumptions you’ve made but haven’t verified? Rework it and try again.

Use case #3: Make a strategic decision

Because gen AI was trained on the knowledge of the internet, it has access to nearly every mental model and decision-making framework ever published. You can ask for general strategy advice, and it will give you a generic way to think about it. But you can also get granular and prompt it to use a specific framework.

How Dan does it:

1. Nail down the decision you need to make. Perhaps you’re trying to figure out how to tackle a bottleneck in your product, or you're not sure what the right next move for your business is. Step 1 is just getting clear on the problem you need to solve.

2. Give AI exact instructions on how to approach the decision. Prompt something like:  “Invert this decision and help me think about the opposite action” or “What mental model would help me think about this decision differently?”

How to take it a step further:

• Save your favorite frameworks to memory. If you feel yourself routinely defaulting to one mental model, add a note in your custom instructions that says “all strategy conversations should reference this model.” The next time you use AI for brainstorming, it will immediately filter its responses through the lens of your preferred framework.
  
  
• Pair mental models with your personal reflections. If you tried out the first use case, translate AI’s analysis into action items. Test it out with prompts like: “You’ve identified XYZ issue that I’m still struggling with – based on my personality, recommend 3 courses of action” or “I set ABC goal for the next 6 months – based on my behavioral patterns, what next steps am I likely to stick to?”

Use case #4: Tackle the blank page in front of you

Dan is a prolific writer, so this is the use case he’s refined the most. In fact, he uses AI for every step of the writing process to help him arrive at a better result.

How Dan does it:

1. Let AI do the grueling research for you. If you’re writing a dense article on philosophy, allow AI to do the background reading. You can do this by prompting it to find and summarize sources – it will also link these for you for quick reference. You can also take a picture of something you’re physically reading and ask it to summarize in whatever style you want.

2. Talk it out with AI and have it give you the TL;DR. ChatGPT’s voice feature is your friend. Go for a walk and just chat through an idea with AI. Then ask it to transcribe the conversation, sift for the highlights, and suggest a topic based on the overall themes.

3. Let it outline your piece with your best thoughts. Sometimes we lose our own plot, which is okay when we have AI. Prompt it with: “Listen to me riff on some thoughts, then put together an outline with the most interesting and relevant thoughts for the point I want to make.”

How to take it a step further:

• Don’t just show it the output you want – give it the input that led to the output. The more reference material you can give AI, the better. For example, if you’re creating a project brief, upload the meeting transcript where you decided on the need for that project brief in addition to an example template.

Use case #5: Build a working app (with no coding background)

As Dan puts it: “English is the new programming language.” He says his most creative (and non-technical) team members are able to stand up decent apps in 2 days with no coding experience by asking AI for help.

How he does it:

1. Scope out your project well. Part of getting usable code is being really clear about what you’re trying to build. Tell AI the purpose of what you’re creating, what coding language it should be in, what devices it’s meant to run on, and everything else you’ve got.

2. Iterate, iterate, iterate. There’s going to be a level of QA to coding with AI. And if you don’t have any programming experience, there will be some confusion and friction to push through. He says those who muscle through it always get there in the end.

How to take it a step further:

• Lean into each LLM’s strengths. With Claude, you can ask it to build a website based on examples, references, and required features. It will code the website, run it, and then give you a sharable link to it. ChatGPT doesn’t have that feature, but it writes code right in your chat. If coding is a weakness of yours, let each LLM use its strengths to augment yours.

Dan’s parting wisdom

We tend to avoid taking on tasks we’re not good at (for Dan, that’s math). But AI gives us the ability to overcome those deficiencies without asking another human for help, whether it’s analyzing data, coding, or writing an email.

‍Want to hand off your weak spots to AI?Become a Section member and get access to unlimited AI courses.