Build the model
Schema, rules, governance, and constraints before screens.
00Archive
I build systems, products, and maps for making ambiguous work easier to understand.
I work close to messy operating systems, name the model underneath them, and build artifacts that help teams see what to do next.
01How I think
Start close to the work, then build the model.
Understand the work as it happens, name the real system, translate it clearly, and build what helps people move.
Model pieces
Get close
Look where the work actually hesitates and routes around.
Translate well
Keep rigor without flattening the people using it.
Use range
Move across domains without losing the structure.
Shorten the loop
Use AI to reach evidence faster without skipping judgment.
02Selected work
The screens are mostly private. The decisions are not.
Public-safe slices of the work: the problem, the bet, the trade-off, and the artifact or outcome that followed.
Artifact stack
Color, as a system
How do you make specialist color judgment usable by thousands of designers without flattening the craft?
Bottoms-up planning
How do you turn bottom-up merchandising planning into a tool teams can use while the operating model is still changing?
Git for non-engineers
Can product managers use a git-native workflow without needing to think like engineers?
Gravatica
What if a company strategy could be read as a source-cited system map instead of a summary?
03Experience overview
Professional terrain, not a prettier timeline.
The useful view is not just where I worked. It is the ambiguity, scale, translation, and judgment the work required.
Terrain map
Enterprise platforms
Internal products where expert workflows, data foundations, APIs, governance, adoption, and executive alignment all have to move together.
Operating-model change
Change work where the product is partly the tool and partly the way teams make decisions, sequence work, and build trust.
Principal Product Manager
Recent work sits where enterprise product systems, AI-enabled prototyping, and operating-model change meet. The useful public story is practical AI adoption inside a large organization, not novelty theater.
Lead Product Manager / Product Director
This work showed the pattern at scale: absorb a specialized domain, build the shared model, ship the platform, and design the adoption path around the humans using it.
04Recent thinking
Notes, frameworks, and product/system reflections.
Writing here is for making the reasoning visible. Current pieces stay honest until they become finished essays.
Reasoning notes
The next interface shift will be navigational
A working thesis on why AI changes how people navigate work, evidence, and decisions, not just how fast they generate artifacts.
Org charts are maps, not hierarchies
How formal structure, influence paths, incentives, and operating cadence combine into the real terrain.
AI will change the shape of product work
A grounded look at what gets automated, what gets amplified, and why product judgment becomes more visible when feedback loops speed up.
05Current inputs
References shaping the work.
Books, publications, tools, and references that keep showing up in the way I think.
Source shelf
Seeing Like a State
A durable reminder that legibility is powerful and dangerous when the map replaces the lived system.
Stripe Press
A reference point for editorial seriousness, software culture, and clear long-form packaging.
Distill
A model for explaining technical ideas with visual reasoning instead of decorative diagrams.
Are.na
A useful pattern for curated knowledge, public inputs, and collections that stay human-scaled.
06Active experiments
Small tests before big claims.
Prototypes, prompts, and unfinished ideas. Some become projects. Some disappear quietly.
Prototype table
Chat as navigation
Can chat become a way to move through a system rather than a box that produces answers?
Model cascade research
Where should lightweight, specialist, and high-reasoning models sit in the same workstream?
Brief to build
How much structure does an agent need before it can build without sanding off the taste?
07Open questions
The archive stays useful by keeping the questions visible.
The work remains active when the unanswered parts stay inspectable.
Open threads
- How can AI shorten feedback loops without outsourcing judgment?
- What makes technical capability become organizational capability?
- How do expert workflows scale without losing the expertise?
- What interfaces help people reason through systems instead of staring at summaries?