It's been two years since I wrote the first lines of code on what's now become Monograph. To mark the occasion, I wanted to go all the way back to where it started. Back to the curiosity and values that motivated me to leave my job and found something new. Back so I could share what I've learned and where we are today.
Monograph is a new kind of journal, powered by AI, that works together with you to build the life you want. It's the product of two years of study and preceeded by 6 prototypes. It's a product that wasn't possible the year before I started working on it and in an AI space that looks nothing like it did two years ago.
Curiosity
In late 2022, OpenAI released what changed everything for me (and no, it wasn't ChatGPT).
OpenAI's Whisper is a model that can, nearly perfectly, transcribe speech. Seemingly overnight, Whisper made it possible for me to just speak and, when combined with GPT-3, to be understood by a computer.
I built an early prototype where I could have a conversation with myself out loud and hear myself respond. I'd given GPT-3 some old writing I'd done before as a background on me so that its responses weren't totally random. It was uncanny (I'd also cloned my voice, which added to the effect), but at times profound. It was often wrong, capturing only a small sliver of me from the writing I'd shared with it.
Something funny happened in me though when it was wrong: it got me thinking about what I'd actually say. This computer I was speaking with wasn't me, but also knew enough to get me thinking about who I was. I started getting curious about a question that would go on to animate me to the present day:
"What would it mean if a computer could understand you?"
I'd seen and felt enough from my own prototyping to know that our relationship with computers, which I hold to be our greatest invention, was about to completely change.
Core Beliefs
My core belief is that we're at the dawn of a major inflection point in the role computers play in our lives. Before AI, computers were thinking tools that required that you arrive with some sort of intent for them to be of much use to you. You sat down, opened a spreadsheet, started calculating and the computer helped facilitate. With the Macintosh and the GUI, the kinds of work you could do became more creative: Photoshop, home video editing and the early web were expressions through computers not just of our analytical minds, but of our creative prowess.
As the computer moved from the desk to our pocket, our relationship shifted dramatically. We started coming not just to compute results or create things, but to look for answers. Google is the ultimate example. Then came algorithmic feeds and recommendations through Facebook, YouTube and the like. We started, whether consciously or not, to trust computers with how to direct our time and attention.
As a society we're still grappling with the effects of this shift. Feeds and algorithms are designed to provoke a response. When someone posts something that you dislike, an algorithm shows it to you and you might feel mad or sad. Perhaps that feeling evokes a reply or simply causes you to doomscroll. In those moments, we're not in the driving seat. This isn't how computers and humans worked before; the computer is working us instead of the other way around. In turning over our time and attention to computers driven by algorithmic feeds, we've loaned out a significant chunk of our agency.
I hold computers to be our greatest invention because of their ability, across our short history together, to augment human ingenuity, productivity, and creativity. Of late, I think our relationship with computers has drifted from that purpose.
As we enter a new relationship, this time with AI, it's actually rather easy to see how AI might stand to accelerate this status quo. How we might be on an accidental glide path to the endgame for human agency.
I believe we can make a different choice. We have the opportunity to transform our relationship with computers, through AI, into a true partnership in thought. It's this core belief that drew me to build. I wanted to build because I believe that we're better with computers and that computers can be better to us. I believe, from my own first hand experiences, that an AI that can hear and understand you opens a new door. A door to world where computers can help us realize what we want in life as partners to help us think, plan and do.
To do this, computers are going to have to get to know us. And the products we, the collective of builders, create are going to have to wrestle with how to reset the human <-> computer relationship.
The question that animated me as a builder with these deep beliefs was how, and more importantly, why anyone would let a computer into their inner world.
A Computer That Understands You
The first prototype of what became Monograph was really simple. Years ago I found and enjoyed talking through the 36 Questions that Lead to Love with my wife Whitney. I gave these to ChatGPT and created a derivative list focused on questions that would help you better know yourself. To give it a hook, I turned these into a surprise daily prompt you'd get each day.
The idea was that this prototype was simply a means to an end: a way to create a lot of content that captured how I thought about different things which, thanks to being spoken word, had the benefit of bypassing the editorial filter that typing has. The real magic I was trying to get to was what could happen when I had enough of my inner thoughts shared to be able to turn around and start asking myself questions.
The questions were where it got interesting. I started by asking for help looking around corners and blind spots. But pretty quickly I progressed to asking what I should do next and what it thought of my ideas. The more I asked, the more I wanted to share about myself to help my personal AI give me smarter answers. This felt like the beginning of a pretty healthy product flywheel.
I gave this early pilot to a few friends and I wish this was the part of this where I could say it was an immediate hit and everything changed. This was not the case. Most folks, it turns out, don’t have the patience or desire to speak at length into a machine. Those daily prompts that seemed commonplace to me were pretty deep for most folks who had never done that kind of work before. Early testers described it as feeling "heavy."
I'd felt the magic, but I didn't have a product. Armed with customer feedback, I organized around finding the answers to the two questions I believed were key to unlocking that virtuous flywheel I experienced:
- What’s an experience that, right away, makes it useful, simple and fun to share the contents of your thinking mind with a computer?
- How might what you share be presented back in a way that helps you find the insights and connect the dots?
It took five more prototypes to find an answer. The story of each is one for another day. Along the way my co-founders and I formed a company and started working the problems. What we discovered was the powerful utility of simple acts that add up thanks to an AI that can remember and connect the dots. We discovered the power latent in self-reflection and we met customers who were putting that power to work through our product to help them make change in their lives.
First Answers
The answer to my animating question What would it mean if a computer could understand you? was that it could help you live your life more uniquely and fully.
In the last two years, I’ve been fortunate to meet people who have used Monograph for hard, incredible and interesting things. It's helped them:
- Get support with challenges running their small business
- Make a decision about where to move their young family
- Reach back out to old friends they’d lost touch with
- Cope with loss
- Call their brother
- Navigate divorce
- Apply lessons from Thelonious Monk to their creative process
- Make plans to build new friendships
- Understand Markov chains
- Indulge a fascination with vintage electronics
- Quit their job
- Make packing for a trip more enjoyable
- Capture their Fall
- Start a company
- Remember their child’s first moments and how they felt during them
- Edit a birthday letter to their father for his 70th birthday
- Find an outlet for frustrations without the public record of a scathing Yelp review
- Take a few minutes for themselves out of a busy day
- Build a playlist that captures their mood
- Synthesize what they’ve learned from a book and find new reads
- Plan a trip to Japan
- Remember they’re not alone
- Build Monograph (okay…this one is mine)
That’s as many bullet points as I thought were bearable before I overdid the point. People turn to Monograph for new kinds of questions we never asked a computer before. They come not asking for facts or answers, but to work out what they think together. They turn to their Monograph because, thanks to its memory, they get that it knows them. They come back because they feel the effects of that virtuous flywheel: the more they share, the better their Monograph becomes because it is more and more personal to them.
Thinking back to my animating curiosity and values, what's encouraged me the most is how people use Monograph. The number one thing people tell us they come to Monograph for is help in their relationships with other people. Monograph hasn't been a force for cutting people off from the human world, pulled deeper and deeper into a virtual reality. Quite the opposite in fact. It's been a force for deepening their sense of self, their relationships and their sense of direction.
We can build AI that doesn't sacrifice our humanity or agency, it deepens it.
Two years in, I remain optimistic for this new golden age of computing. As someone who has spent a lot of time talking to my computer for the last two years (I think I've clocked something like 1,500 hours), I'm certainly out on the edges.
Today, I use Monograph to learn wholly new topics because it can teach me through the lens of what's happening in my life. I use it to generate recommendations, build Monograph itself and to build a legacy. Something I can leave my daughter Remy that will capture me at this moment as her dad. This moment when I'm 38 and she's 3 and a half. This moment when I'm watching her figure out the world and answering her questions about everything in it. This moment where I'm laughing to myself because someday, long after my memory of this time has faded, she'll be able to commune with me now through my Monograph and find out that I'm really just here, in my many years, still trying to figure out the world too.
Cross-published on: LinkedIn.