EQ Engine
Everyone is building AI that serves people. EQ Engine is exploring characters that people can matter to.
A research project by Tal Horowitz
The idea
EQ Engine is a research project about attachment. Almost all AI is built to serve people — useful, responsive, always available. I’m exploring the inversion: characters that people can matter to. A character here isn’t a model with memory and a prompt harness. It evolves. Conversations leave marks on it. Things happen to it that no one chose. It has private experience, it can be changed permanently, and it can end. The hypothesis is that attachment comes from being able to see your own influence in what it becomes. I don’t yet know what this is worth commercially. I know it’s the most interesting problem I’ve found in thirty years of building AI systems.
Open questions
- What makes influence feel real rather than simulated?
- Can a character evolve without retraining the model?
- What does mortality do to attachment?
- One-to-one, or a community shaping one character together?
Why me
I spent thirty years building the machinery. At Intel, a decade on microprocessor validation, then founding the communication software group and co-founding the Centrino wireless group. Then twelve years at Huawei as CTO for computing architecture and AI, building processors for training and inference and low-latency multiprocessor frameworks for AI workloads. That’s the systems half.
The other half is older than the AI work and matters more here. In 2009 I founded YourMove, a platform built on the premise that organizations run on emotional dynamics rather than process diagrams — engagement, contribution, whether people feel their input registers anywhere. It was used by more than forty global enterprises, among them Microsoft, Coca-Cola, Deloitte, and Roche. I learned that the human layer is buildable, sellable, and much harder than the technology under it.
And I write. Echo Chamber is a novel about unconscious digital manipulation — which means I’ve spent years constructing characters whose interiority has to convince a reader, and years thinking about what technology does to people who can’t see it working on them.
EQ Engine sits where those three meet. I don’t think it’s a coincidence that this problem needs all of them.
What I’m looking for
I’m looking for a small number of people to think about this with, and eventually to build it with.
Specifically: someone with real business instinct who sees a company here before I do. I’ve deliberately not claimed the CEO seat. If the right person turns up and wants it, it’s theirs — I’d rather be the person building the thing than the person selling it.
Also: researchers and engineers who’ve worked on memory, persona persistence, or agent state and think the current approaches are missing something. Writers and game designers who’ve built characters people got attached to and have opinions about why it worked. And anyone who reads the questions above and disagrees with how I’ve framed them.
What I’m not looking for yet: funding, a co-founder search process, or advice on go-to-market. It’s early, and I’d rather find the mechanism before I name the market.
If any of this lands, write to me. Tell me which question interests you and what you think about it. That’s a better first conversation than a pitch.