Active and evolving
Jose AI Assistant
How do you show how someone thinks before working with them?
The hard part wasn’t the model. It was organizing experience into a representation I could trust.
PRODUCT LEADER · APPLIED AI · DECISION SYSTEMS
Complex models become workflows people trust.
Across more than 15 years in finance and product, my work has centered on one problem: helping people make better decisions in complex systems.
I spend my weekdays building AI products and my weekends riding my motorcycle, surfing, or helping keep a 2009 Jetta alive. In between, I’m usually reading about AI, economics, or whatever has captured my curiosity that week—ideally over coffee or mate.
Jose AI Assistant
Explore the work
Ask about Jose’s experience, projects, decisions, and ideas. Every answer stays grounded in evidence from this site.
Grounded in the source material behind this site. If an answer matters, follow it back to the work.How I tested itPrivacy note
Projects & Essays
What I’ve built and what I’m still exploring.
Over the past year, many of these projects have revolved around a simple question: how do I make better decisions? That led me from reflection systems, to personal knowledge, to AI interfaces that help organize and challenge my thinking.
Completed analysis
HomeTap Analysis
Where does value actually come from in a home-equity investment product?
Understanding the customer wasn’t enough; I had to model the economics their decisions created.
Active and evolving
Compass
How do you keep one bad day from becoming a bad week?
Reflection creates better feedback than a perfect checklist.
Built and iterated
Training System
Can I use GPT as my workout coach without trusting it to remember my workout history?
Reasoning belongs in the model. Memory belongs in software.
Evolving
Agent Workspace
Can agents help build and update a website faster while preserving the judgment behind it?
Still emerging: agents may make work faster, more accurate, and more satisfying when each one has a clear role and review loop.
Writing
The Model Is Not the Product
Why do so many impressive AI demos fail to become great products?
The technology changed. The product problem didn’t.
Launched small and iterating
Record Library
Could a music app optimize for memories instead of engagement?
Launching small taught me faster than polishing alone. The project started for me, then began becoming something someone else might love.
Built and used
The Daily
Can I design the perfect day?
I didn’t discover the perfect checklist. I discovered the limits of one.
Evolving
Capture
Can a knowledge system surface the right idea at the right moment?
Capturing information is easy. The hard part is knowing what matters—and when it matters.
Lately,September 2026
Occupying my thoughts
An obsession is free motivation. If it’s pointed at something worthwhile, pursue it to the max. Don’t seek balance. It will become part of your identity.
Reading
Books shaping my thinking
The IliadHomer
Project Hail MaryAndy Weir
Dopamine NationAnna Lembke
The Man Who Solved the MarketGregory Zuckerman
Watching
Ideas I’m studying
Human Behavioral BiologyRobert Sapolsky
Thinking / Writing
Ideas I’m developing
The Model Is Not the Product
How AI reshapes roles and expectations
Building / Doing
Experiments I’m running
CaptureA knowledge system for surfacing the right idea at the right moment
CompassA personal operating system for reflection and action
Also, lately: trying to perfect my bolognese sauce.
Journey
The thread
Helping people make better decisions in complex systems.
2005–2014
Learning to Think in Models
Universidad Torcuato Di Tella • B.A. in Business Economics • Master in Finance
Economics taught me to think in incentives and constraints; finance taught me to quantify uncertainty and use models without mistaking them for judgment.
2014–2017
Learning to Invest Under Uncertainty
Bank of America Merrill Lynch teams • TPCG
Investing taught me to make decisions with incomplete information, separate signal from noise, and turn complexity into action.
2017–2025
Learning to Build Products
Duke University MBA • Zest AI
Building AI products taught me that accurate models are not enough—the real product is the decision system around them.
2025–Present
Building My Own Convictions
Independent • Compass • RecordLibrary • Writing • AI systems
Stepping away clarified the through-line: helping people make better decisions in complex systems, and making that judgment visible through building and writing.
How I Operate
What I return to
Principles for building, deciding, and working with others.
- 01
Build My Own Mental Model
I don’t like making recommendations based on borrowed understanding.
- 02
Quantify the Tradeoffs
Every meaningful decision involves tradeoffs. I make them explicit whenever possible.
- 03
Design with Intention
Every feature, workflow, and intervention should have a clear reason to exist.
- 04
Understand People
Technology alone does not create change. People do.
- 05
Build Shared Understanding
Complex work moves faster when people share the same representation of the problem.