Azeem Azhar asked members of his community for their current favorite ways to use various AI tools, and I thought I’d share my current list publicly here. I could go on, but perhaps I should write something like this quarterly. I love experimenting with things like this and I bring almost all of these into every consulting engagement I do, too.

Here are short descriptions, followed by the specific AI that I find most effective for each of these use cases.

• “What have I written in my notes over the years that’s most relevant to the following topic?” This is my #1 go to for things both fun and serious. (Claude Project, described here)
• Asking for counter-evidence to any text I’m reading. Also really helps me understand the original text more quickly. That recent Nature article on how AI is essential to solving climate change? ChatGPT took some steam out of that for sure. (ChatGPT with a javascript bookmarklet, described here)
• Create a set of synthetic personas that are relevant to a given topic and have them answer a question I have, thinking through step-by-step. The answer can be open ended on a topic, or specific like “what do you think of this email?” I do this every day and it always brings up points of view I am thankful to consider. (Claude)
• Make a briefing book about a person or an organization I’m about to meet. (Perplexity) Now make a personalized version of it for a person I’m going to introduce the first person to, highlighting similarities and differences between their work (Perplexity) and then turn that briefing book into an attractive one page artifact (Claude)
• Perform a combinatorial analysis of the X! ways that two sets of information could cross-pollinate; for example, consider a big set of updates from a set of organizations in a network, and ask “what could be done if update #1 were connected to update #2, etc?” (Google NotebookLM) Then score those combinatorial possibilities based on a set of criteria like “would this combination have transformative potential?” “Are these organizations geographically near each other?” And perhaps most importantly “is there a discoverable history of these organizations working together previously?” The combinations that score Yes, Yes, No to those three questions are rich opportunities. (Perplexity, see screenshot below for example)
• Summarize (for example) this email newsletter in an even more succinct fashion. (Comet, the AI browser from Perplexity)
• Out of this set of 1,000 updates from organizations in my business sector – which 10 should I highlight for the following email subscriber, and why? (Claude by API, followed by a mailmerge. See my green tech and sustainability newsletter, for example. Every subscriber gets their own personalized version.)

And I’ll sneak one more in: I often ask any of the AIs I’m working with what Azeem Azhar might say about a topic I’m thinking about. (I also ask about adrienne maree brown‘s POV.) Azeem has published so much online that all the AIs have a pretty good body of text to work with. And I really appreciate his… exponential view on things. Give it a try yourself!

Below: A slide I drafted based on the combinatorial analysis example below.

The holy grail of AI for many of us is to chat with the notes we’ve taken over the years, isn’t it?

If you read a lot, and write down a lot, you can probably see (always just over the horizon) the potential power of AI to help unlock incredible value hiding inside our archive of notes.

At this point there are probably multiple tech solutions that are good enough from a tech perspective: Custom GPTs, Google Notebook LM, Notion, various startups. The remaining challenge is in human workflow.

I’ve been experimenting lately with a workflow and tech combination that is SO EXCITING to me that I wanted to write a blog post about it. And incidentally, when I work with organizations I always bring years of knowledge and experience to the table – but this feels like a big step up in how powerfully I’m able to do that.

Claude Projects + Obsidian + a Stopwatch

1. Each morning I start a timer for 5 minutes
2. Then I open Claude and go to a project called My Recent Notes. I click to add new content to the Project Knowledge section.
3. Then I open my note taking app of choice (it’s Obsidian) and I navigate to yesterday’s notes. Select all, copy, paste into Claude.
4. That takes 30 seconds, so now I spend the remaining 4:30 clicking the Random Note button in Obsidian and just bouncing around through years of notes archives I have, opening pages, select all, copy, paste into Claude.

Ta da! That’s it. After the first day or two, this was already my #1 favorite place to chat.

Is there a way to turn the whole markdown file of my years of Claude notes into one file I can upload into Claude? Put it all in Google Drive and let it get updated automatically? I don’t know. I’ve tried spending some time exploring options like that and honestly, I’d rather just spend 5 minutes a day re-visiting my notes – yesterday’s and random ones, and grow this “organically.” It’s already super useful.

A few tips for getting the most out of it:

1. Pro and con notes: I like just having normal chats there, after I edited the system instruction to say “please refer to specific notes wherever possible,” but I also really like asking two questions:
   • Which notes in this collection are most relevant to the following thesis?
   • Which notes in this collection contradict the following thesis? (this is a really good one)
2. One of several chats: Yesterday I had one tab open where I was asking my recent notes what they could contribute to a question I was wrestling with, but while that tab was thinking – I copied my prompt, opened up a second Claude tab not querying my Project and asked the same question there. Then a third tab to ask the same question of ChatGPT. In a quick minute, I had three distinct takes on my question. I quickly scanned over all three, picked out what I liked from each, and created an amalgamated answer to my question with pen and paper. And I love where we landed!
   

I’ve tried a lot of things over the years and I’ve talked to a lot of people who have wanted something like this. The above system so far is fast, cheap, and good enough to prove very exciting to me.

It gets me excited about note taking again too, which is exciting.

You might also like my friend Alexandra Samuel’s approach, which she wrote about this week in her newsletter edition Give AI the Knowledge it Needs.

I am always struggling with the relationship between aspirations and capacity, something I’ve really only grown aware of in recent years.  In part by blowing far beyond even my substantial capacity for too long with some family crises, and in part by reading the really smart book On Grand Strategy, which is largely focused on the singular point that aspirations must not exceed capacity.

One of the ways I try to get more done inside the time, space, and energy I do have is to try to spend 5 minutes doing a lot of things.  Can I do that thing in five minutes?  Could a five minute version of that thing be good enough?  Very often the answer is yes.

I also turn it around and imagine my future self saying to my present self, “you couldn’t even spend 5 minutes on that??”  I don’t want that to happen.

One of the most productive things I do almost every day is sit and think through a single topic for five minutes, usually with pen and paper.  I will also read a book for 5 minutes if that’s all I can do.

You know the GTD advice that if something can be done in 2 minutes, then you should do it right away? I think this is a related concept.

And that’s five minutes of blogging on that.

David Gurteen talks about truncated “Knowledge Cafes” with five minute talks.  That reminds me that I learned how much could be done in five minutes when I was in high school and won many speech and debate tournaments in Impromptu Speaking.  In that event, you’d be given three philosophical quotes, pick one, take 30 seconds to prep, and then give a 5 minute speech about it.  I would apply the perspectives of 3 different interesting thinkers or other people in history to the quote I’d picked, analyzing it from 3 different perspectives.  That was probably where I learned too about Symphonic Thinking, Daniel Pink’s term for the ability to generate connections between seemingly disconnected things.

And that’s five more minutes.  Cheating? Maybe.  Now if you’ll excuse me, I’m off to spend 5 minutes thinking through an important life matter.

A friend asked me recently what some of the core principles are in my note taking and review system. I get a whole lot of value out of my note system and I love talking to notes nerds.  But not notes for notes’ sake! For making an impact on the world, for the better.

When I read Dan Pink’s book A Whole New Mind and learned about what he calls Symphonic Thinking, or the ability to find connections between seemingly disparate entities, as a key thinking pattern for the future of work, I thought “wow, that’s what I’ve been doing already! I’m going to do it more deliberately!”  And so I regularly cite research, reading, things I’ve heard on podcasts and more in my day job and my work outside of work.  It’s one of my superpowers, but I really believe it’s something far more people could help the world with. I recently changed my Twitter bio to read: “Sharing thoughts for growth-oriented people about how information can be synthesized to build power to make the world a better place.”

My notes come from a wide variety of sources, but most commonly from things I’ve stopped and typed up into Roam Research after I’ve heard them read aloud to me either by Pocket, which I feed with links from Twitter using an IFTT applet that sends the links in any tweets I favorite to pocket, as well as a few key RSS feeds. I also read a lot of PDFs by text to speech using hte Voice Aloud PDF to speech iOS app on my phone. I “read” with my ears and clean my kitchen a lot, or jog. I’ve also been experimenting lately with spending 5 minutes scanning through Feedly RSS reader to find things to toss into Pocket and listen to read aloud.

So I read a lot (I also miss a lot, it’s ok) and then I stop when I hear something really good and I write it down in Roam. (Or if I’m jogging, I associate each thing I want to remember with one of my limbs, then I go through them one at a time “left arm, left leg…” when I’m done running and I write them down.)

These are some of the things that came to mind when my friend asked about my note taking and review system.

• Make it easy to take notes when you can. The Roam unofficial mobile interface has been essential for me.
• Failure to cite your sources can be a real pain – finding an easy and repeatable way to cite where a note came from can make a big difference. I have notes from years ago where the insight is good but I didn’t record the source and it’s a real bummer.
• Make it easy to review your notes, Anki flashcards are super helpful. Anki tells me I’ve been adding the equivalent of one card per day to my “lessons being learned” deck since late 2015.
• Make it easy to recall half-remembered flashcards, Roam’s search or my personal wiki control-F have been really helpful. I search in Anki sometimes too.  I regularly have a hazy memory of something but search can bring it up in the background while I’m on a call for work.
• Review as much as you can, as often as you can. It’s just like an athlete practicing. I try to spend 10 or 15 minutes on my flashcards each day. I wish I spent more time, I’d love to double that. Or more.  That’s something I’m actively working on right now.  You can see my scorecard there on the right. Lots of room for improvement, but that’s better than a poke in the eye. If I could consistently review 20 flashcards a day and really integrate the acquired wisdom with my life…that would be amazing.
• As I review the flashcards, I try to think about a real life scenario I could apply that concept to. That makes it far more real and stick in my mind.
• After each review session, there’s often one flashcard in particular that I really dig into. I might spend 5 minutes writing about it. I might just think about it while going about my day.

Those are some of the things I’ve been doing for the past few years. I’d love to hear anyone else’s tips and tricks you use as well. Ultimately, I think a lot of it is just about showing up. Being imperfect, coming back to the path, and applying what you’ve learned and reviewed in the real world.

Below, some examples of 3 of my most recently added flashcards.  These ones Anki is going to show me again in a couple of days, but in time it will space them out over years.  Good luck to you in your studies and their application!

There’s a saying that simple decisions are best made with rational thought alone, but complex decisions benefit from a big dose of gut feeling as well.

I’ve been employing two methods for dealing with both types of decisions that I thought I’d share here.  I think of things like these as tools I can learn, practice with, get better at, and then deploy in my work.  I typically pick them up reading online, record them on my personal wiki page for reading notes, then transfer them monthly into a variable repetition mobile flashcard app where I review and learn them over time.

For simple, but hard, decisions, I’ve been using for several years a method I call “write it all down and pick 6.”  I learned it in a print edition of HBR that I picked up at an airport , I think it was Stanford’s Baba Shiv that suggested it but to be honest I didn’t write that part down!

The idea is: write down all the factors to take into consideration in your decision.  As many as you can possibly think of.  This feels great, like you’ve really given it a good thought.  Then, pick a very small number of those factors that are most important – at most 6.  Now look just at those 6 most important factors and honestly ask yourself what decision they support making.  This may be more powerful than it sounds.  It’s great.

Second, when you’ve got a complex decision, it can be helpful to break it down into simpler considerations first. “Simple is about relationships between individual people, objects, beings, truths,” writes Adrienne Marie Brown in her wonderful recent book Emergent Strategy.  “Part of what can clear a path to making things easier is to name the simple interactions at play in a complex system.”  Brown credits  Rachel Plattus of Beautiful Solutions with that insight.

That’s a new model I’m going to try out. If it’s a simple decision, use Pick 6.  If it’s a complex decision, try to create some simplicity by thinking about constituent relationships, then pick 6, then add gut feeling.  Gut feeling is high-bandwidth, high-context, powerful stuff too.

Related: The Conscious Competence model. When you start learning a new skill you are unconsciously incompetent, then consciously incompetent, then consciously competent with practice, until ultimately you can be unconsciously competent. Competent with far less effort.  That’s from Ed Batista’s blog post on Authenticity and Leadership.  Ed credits Martin Broadwell, a Bible teacher in Decatur, Georgia, who developed the Conscious Competence model in 1969.

Please share any related thoughts.  Thanks!