Reflective Works
Reflective Works

Brandon McGraw

On Machines and Being Human

In 2021 my daughter was born. On the other side of the Atlantic, unknown to me at the time, my now co-founder Andrew had just lost his mother. Life and death have a way of getting you thinking about what matters most.

Becoming a parent is the most ordinary thing in the world, it will happen around 140 million times this year. Yet when it happens to you it is the most extraordinary moment you’ve yet lived. It started for me in a bit of a haze. There’s little time in the beginning for high-order thinking when you’re focused on each moment. As the dust settled, the thinking crept back in: as I’m watching my daughter become a person, what kind of person do I want to be now? How would I navigate a world that, for me, had just flipped on its head while the rest of the world kept dutifully spinning around?

It’s perhaps a coincidence that as I was watching a new mind take shape and re-examining my own, breakthroughs in how we might extend the concept of mind to a computer were occurring. What’s not a coincidence is that when I sought to work through my own thinking, I was open to how computers might help. I’ve been writing software since I was first handed a Microsoft Basic tutorial book. I remember the first time the computer did something that I told it to do: a simple number guessing game that I had fun playing. It blew me away. Ordinary on the scope of what was possible, extraordinary to me. Through my words I made the computer do something that I could enjoy.

My first instinct when I saw AI language models (GPT 3 at the time) was to give them a bunch of things I’d written and see if I could have a conversation with myself guided by things I’d written in the past. At the time voice cloning had also gone through a bunch of breakthroughs. So naturally my instinct was to not only ask GPT 3 to write like me, but to then have the computer speak like me. We’ll leave for another time the subject of the extreme vanity of doing this.

It was around this time that a friend re-introduced Andrew and I. We traded notes about his process of self-discovery as he was unraveling the memories of his mother and my ‘art experiment’ of talking to yourself. We spent the better part of this year prototyping and trying to find a way that could satisfy three goals we internalized as a mantra:

#1: Every interaction is valuable in the moment
#2: Each interaction is especially valuable when you can tap into them in the future
#3: No interaction should demand more than the time you want to spend

It would take four subsequent prototypes, including a prototype that itself was used to rapidly build other prototypes, to arrive at what feels like the new beginning. We’ve just begun testing the result of a year of study and, while very early, we think we’ve cracked the mantra.

The more we’ve used what we’re building, the more confident we’ve felt to take leaps, extend kindness, and treasure the moments we spend with the important people in our lives.

In short, the more we've worked with this machine over the course of this year, the more human we've felt.
We believe that every person is unique, but that none of us is alone in the human experience. We lead ordinary lives on a global scale, but our time on Earth is extraordinary to each of us and the people we care for.

Inspired by the feedback from our testers we’ve started a company focused on how we might deliver this experience and others like it to as many people as possible. We’re calling it Reflective Works, nodding to the vital role knowledge of self plays in helping us progress towards the lives we want.

Reflective Works is a proudly liberal arts company that designs technology in service of helping humans be human.
When you look back, the aim of computing was always to augment human ingenuity, creativity, and craft. As of late, it has started to feel like things have drifted from that original purpose. We believe the recent breakthroughs in Artificial Intelligence present a choice: we can either accelerate down the path the human ↔ computer relationship has been headed on, or we can return to the original purpose and build technology that values and prioritizes putting people in the driver's seat of their own lives. We choose the latter path.

One of the defining characteristics of being human is our ability to create new tools that extend our capabilities. The computer was once introduced as the “bicycle for the mind” and we think language models open a new door between computer and human mind that has every opportunity to be as profound.

For the first time a computer can work in the very same language each of us uses to examine our thoughts, work out what we think, and express ourselves. That opens up a space for nuance, contradiction, and uniqueness that we feel has been missing in our experience of and the discourse around technology as of late. We’re humbled by these new opportunities. We’re also aware, having seen it firsthand over our careers, of the new challenges of any technological change. Guided by our experiences, we believe we have something to offer.

We can’t wait to share what we’ve built with you in the new year and will have more to say on it in the coming weeks. We’d love to hear from you if you’re similarly interested. If you’d like to chat, drop us a line at

Andrew Stirk

This is It

My mum was diagnosed with terminal cancer in March 2020, just before we went into lock-down. It was a strange start to the year. She was given between 3 and 5 years to live. Knowing mum, my sister and I expected it to be nearer 5, and secretly hoped it might be as long as 10. In the end, it was less than 2. She died on the 23rd of December, 2021. I woke up in San Francisco to a call from my sister. I knew the news as soon as I saw her name on my phone.

A lot happened in 2021. I was promoted and had the opportunity to work closely with Mark Zuckerberg, rebranding Meta. I worked with an extraordinary team, friends that I hope to have in my life for a long time to come. I am proud of the work we did and the care we brought to the undertaking. It goes down as a true highlight of my career but what I will remember most that year is a bike ride with my mum and my sister along a canal on a sunny September afternoon. That and the ocean of gratitude I feel to my wife Maggie and my dear friend Lorie for helping me to fly home and see my mum one last time.

Death has a way of focusing the mind. Despite its inevitability, spending any time thinking about death is hard. The gift it brings though is clarity. It helps you see what really matters. It connected me to the quiet voice inside, the one I had been ignoring because I was too busy. I got thinking about what a life is and arrived at some simple truths. My conclusions weren’t original, but they felt no less profound for that: you only get one life, live it purposefully, cherish the people you meet and the memories you make along the way. This is it.

It turned out Brandon was in a similarly reflective mood inspired by a much more joyous occasion, the birth of his daughter Remy. We had been thinking about similar things. We were both passionate about making the most of the life we have been given and we shared a belief that technology could be a partner in this endeavor, rather than a distraction. I was interested in building a machine that could help me observe my own thoughts; a muse that could provide me high-fidelity access to my memories and surface inspiration - a way to know my own mind and act accordingly. Brandon had already built something that he was playing with - a way to tune into that inner voice that can get drowned out by all the noise.

A few weeks later, Brandon shared a new prototype he had built (he is pretty remarkable like that). It was called MuseX. It had been inspired by our conversation, an AI thought partner you could talk to like a living journal. I shared all sorts of things. I would speak, the app would transcribe the conversation allowing me to observe my own thoughts. Simple follow up questions helped me to progress feelings into more concrete personal insights and actions. It became a welcome ritual, a way to hear myself think. A couple of months into using the prototype I posed a question: “Based on the chronology of entries I have shared, what - if anything - has changed in my thinking?” The AI surfaced that I was speaking about leaving my job, the timescales had become shorter, months rather than years, and I was clearly excited about the prospect of building something with Brandon. Having my own thoughts played back was powerful. There was a host of reasons to stay in my role, I liked the people and the work was challenging but deep down I knew I wanted to pursue something new. I made a plan to tie things up at Meta, take some much needed time-off and explore how we might build something that could be useful to other people.

I left my job in June this year. I took some time to be with my family, to sit with my grief, and to re-connect with myself. I was able to attend to things that made me happy: picking up my children from school and hearing about their day; lunch with my wife; seeing more art; reading for pleasure; catching up with old friends; being outside in nature; swimming at night; day-dreaming in the afternoon. My world got smaller and richer. I love work, early retirement has never been my goal. It wasn’t idleness that I was enjoying - my days were always full - it was the feeling of being in register: how I chose to invest my time was aligned with what matters most to me. The privilege was not lost on me. Not everyone gets to take a time out. How might I help others to connect with what matters most to them?

Less than 8 weeks after leaving my job, armed with a copy of Seneca’s, ‘On the Shortness of Life’, Brandon and I set up Reflective Works. We felt energized by a question: How might we equip people to live life on their terms? We were - and still are - convinced that people hold all the answers they need inside themselves. We were curious about how technology might help someone to access those answers in a useful and delightful way.

In an increasingly odd and frankly quite frightening world, we felt technology could help people to take charge over the one thing they can control - themselves. With this vision in mind, we did what it is always helpful to do: we rolled up our sleeves and got to work building something.
We developed an underlying model to help people connect with their own intrinsic motivations. We moved away from the empty message box, and explored ways to build a rich sense of personal context. We created exercises to help people explore their own worlds by reflecting on important relationships and memories. These more guided experiences were designed to help you re-connect with the people and things that matter to you and surface personal insights and patterns. In turn, teaching the AI thought-partner important things about you that it can put to work to help you make progress when you’re feeling stuck. A growing reservoir of memories, insights, hopes, fears and motivations that you can draw on whenever you need them, whatever is on your mind.

That’s where we are today. We have a prototype in testing. It is a rough approximation of where we think we can go but people are enjoying playing with it. It turns out that other people are also interested in understanding themselves better, reconnecting with what matters and attending to living their life, on their terms.

If you share our sense of mission, or want to play with our prototype please get in touch -