I started programming the flocking of birds in 2012 where I saw that simple rules followed by each bird - to not collide with their fellow, to not stray too far, and to align with those near them - produced incredibly complex and beautiful emergent behavior that is impossible to predict from the top down.

I spent years working with language as code, and when I started working with LLMs in 2019 (then GPT-2) I saw that similar to the steering vectors that described the flight of those birds, that these language machines designed to predict the next word also had a kind of velocity, and that they were steered by narrative - the directional vector of meaning beneath statements that say something true far beyond their single sentence.

As Chief Strategist at Vouch.io, I developed a manual process for steering LLMs with narrative. Instead of trying to train new models, I treated the models themselves like a computer, and a narrative architecture - an interconnected web of narratives that define a way of thinking - like a program that can be installed into them. We download how an organization, project, or individual views the world - the narratives that guide their next thoughts and ideas. I used this to bootstrap any AI to instantly think like the company: to create content, test ideas, even write code. But instead of defaulting to the average of the Internet, the underlying model was already flying in the direction of the company.

I left Vouch in September 2025 to automate the process and built Aswritten.ai — collective memory for AI. Every conversation with a coworker, every piece of client feedback, every strategic argument, every engineering decision — these are all course corrections that currently evaporate. Aswritten catches the heading change before it’s lost and breaks it down into a narrative architecture of steering statements that point toward where we’re going, not just where we are today.

Each memory represents a shift in narrative. We are so used to working on content itself, but what we are really working on is our ideas, our vision; there is a place we are heading together and my work has been to create a process that takes new ideas as new directions that change our course.

When we create narrative as code, we can now treat the way you speak, the choices you make, and the visions you share together as code as well. We can collaborate, version, manipulate, review and deploy perspective, as written.

LLMs become not just a tool for conversation or coding, but a compiler that takes a narrative architecture and produces a synthetic individual whose every thought projects far further down the paths we set than we could tread alone.

We too are compiled, from stories inherited and newly written. And it is the shifts that enrapture us, the conflicts, the transitions, the causes.

I have seen the incredible productivity that LLMs have given us. But they have turned our starting line from the blank page to the complete and misguided draft. I built aswritten to feather our arrows, that our first attempt may always let us fly further in the direction we’re headed. That our work is to adjust our aim so that all arrows find target, not spend our time moving the ones that already missed.

Our work together, as written.