How I use AI to solve (almost) any problem

Part III of The AI skeptic’s guide to AI collaboration

Hello friends! Today I want to share my AI-powered approach to solving problems. I’ve used this process recently on everything from setting team success metrics to finding a new hand soap after Muji discontinued the only hand soap I have ever truly loved. No matter the multitudes you contain, I’m here to help you figure out a path forward.

But first!! I’m excited to share that I’m partnering with Maven to teach a course on becoming a supermanager with AI. My recent post on Lenny's Newsletter about using AI to scale yourself as a leader is now his 7th most popular post of all time (🤯), and I’ve heard from tons of people interested in learning more about my approach. If you’re interested in building AI tools that help you become a more effective manager, please take this survey here to help me shape the course content.

Now, back to our task at hand. You may have noticed that my post last week on writing with AI was as much about my writing process as it was about AI. Let’s unpack what this means for solving problems.

The general structure I follow when solving a problem is:

1. Define the problem and get clear what success looks like—like to the level of writing out success criteria

2. Brainstorm solutions and assess them against the criteria to quickly prioritize

3. Develop a plan

4. Validate!

Again, this same basic process works regardless of the scale of the problem. I might use it to:

1. Get my email or calendar under control

2. Build a new healthy routine or figure out how to actually take care of my cuticles

3. Figure out the right strategic priorities for my team

4. Navigate a challenging interpersonal situation

5. Fix a broken process at work

6. YES, FIND NEW HAND SOAP……I realize this is insane but it was fun tbh

As an example, let’s use start a problem that I hope lots of people are thinking about as we embark on the new year: how can I take more vacation in 2025? While AI may be an overpowered solution for this, I think it helpfully illustrates the steps in an accessible way.

First, load up your AI tool of choice. I go back and forth between ChatGPT 4o and Claude 3.5 Sonnet depending on whether I am feeling more serious and logical (ChatGPT) or more creative and thoughtful (Claude).

1. Frame the problem and give context: Take time to thoroughly describe your situation. What's frustrating you? What have you already tried? The more context you can provide upfront, the better AI can help structure your approach.

   1. Prompt: I want to figure out why I’m not taking enough vacation. Here's what I know so far: I always think about planning a trip but end up overwhelmed by all the options. When I do try to plan, I feel guilty about taking time off because I worry about work piling up. I’ve also realized I don’t actually block out time to plan vacations in my calendar. Can you help me identify what’s stopping me from taking vacations and suggest ways to address these challenges?

2. Define success criteria:If you’re overwhelmed by options, it’s hard to get started. AI can help you focus by refining what you’re really looking for. You might already have a vague idea of the kind of vacation you enjoy—taking a few minutes to clarify and define success criteria will make AI infinitely more helpful.

   1. Prompt: I heard about this resort in Fiji that looks like my ideal vacation destination, but I want to go somewhere closer to home. Let me share some information about it, and then I want you to tell me some specific words that you would use to describe what makes it unique and what type of vacation you would expect I would have there.

   2. The AI might respond with things like:

      1. Tranquil – Capturing the peaceful, serene atmosphere of the resort.

      2. Immersive – Referring to the rich cultural and natural experiences offered.

      3. Intimate – Describing the personalized service and cozy accommodations.

      4. Adventurous – Highlighting activities like diving, snorkeling, and exploring the reef

      5. Eco-luxe – Balancing sustainable living with comfort and elegance.

   3. Prompt: Great, let’s go with [choose the words that resonate]. I want to find an option that matches that vibe, that is also less than an 8 hour plane ride from Boston, and which will cost less than $x/night. Before we do that, can you help me set evaluation criteria based on this that will help me rate options?

3. Generate Solutions and prioritize. Now that you have a framework in place, ask the AI to come up with solutions (in this case, locations) that score high on that criteria.

   1. Prompt: Great. can you recommend 20 countries or states that have accommodations that will score highly against those criteria?

   2. Prompt: Please score each of these options against the success criteria and highlight the top 3.

4. Iterate. Does it feel like the AI is missing something? Maybe you forgot to mention that you’d like to go somewhere warm in February, or that you really want access to the water. Give it feedback and keep iterating until you land on something you like.

5. Make a plan. I like to start by finding accommodations I’m excited about, then planning an itinerary around those. So I might take the top 5 places the AI suggested I go, and ask:

   1. Prompt: Let’s go with your top 5 recommendations. For each of those, can you recommend five accommodations that also fit my criteria?

   2. Prompt: Can you score each of those accommodations against my success criteria and explain why you gave it that score?

   3. Prompt: Zoom out for a second and forget the success criteria. Knowing what you know about me, what do you recommend I do?

6. Make a plan….in detail! You can go on as long as you like here. Do some of your own research, come back to the AI, go back on your own, whatever works.

   1. Prompt: Can you recommend an itinerary for me if I stay at x resort that is not the obvious itinerary that everyone does?

   2. Prompt: What are some things I might enjoy here that I might not learn about in a guidebook?

   3. Prompt: How can I best show my respect and appreciation for the culture in x country?

7. Validate. Remember, the AI is here to help, but you’re still driving. I always encourage people to get another set of eyes on your work, too.

And that’s it! It’s really so simple. Just:

1. Tell the AI your problem and what you have already tried.

2. Describe a successful outcome and ask for help turning that into success criteria

3. Ask for solutions that align with your vision for success, then prioritize

4. Ask it to make a plan for you. In this case it was pretty simple but for a more complex problem you might say, “Suggest an extremely specific step-by-step plan for me to resolve this problem.”

5. Keep the humans in the drivers seat!

Random Recs:

1. Surely at least one person is wondering what hand soap I ended up with. The answer is Snowdrop by Mrs Meyers. I do not love it as much as my old Muji soap but it will suffice. My friend Nejla who is a scent expert smelled it and said “that’s good stuff.” So I’ll count that as a win, for now. My success criteria….

   1. Scent Quality

   2. Moisturizing Ability

   3. Ingredient Quality

   4. Price and Value - yes I saw that tomato hand soap in all the gift guides and yes it appeals to me greatly but I CANNOOTTTT bring myself to pay that much for soap without feeling like I have totally lost the plot.

2. My definitely too late, single item gift guide: this marbled enamel 1 qt colander may be my favorite consumer purchase of the year. It is so cute and small and the perfect size for rinsing one serving of berries at a time so the others don’t get moldy. The perfect gift for the person in your life who loves to rinse one serving of berries at a time to put on her oatmeal and will also not mind if your gift arrives late.

Thanks for reading and, again, please take this survey here if you’d be interested in joining a course on using AI to scale yourself as a manager. It’ll be fun!!