Home

Thinking

The Pattern Underneath

Principles and tools

The things I work on can look very different. The way I approach them usually isn't.

I look for what people actually do, make the assumptions visible, test the model before scaling it, and build systems that make the work more repeatable.

These are the principles and tools behind that process.

01

Principles

01 Build the system

Not just the solution.

Solving a problem once is useful. Understanding why it happened and building a system that handles it repeatedly is more valuable.

I look for the process underneath the immediate problem: what information needs to move, who needs to make a decision, where failure can occur, and what should become repeatable.

Enterprise Model Execution Platform: standardized model integration instead of solving each onboarding independently.

Nêm Nếm: turned recurring event preparation into an operating system rather than rebuilding the process for every dinner.

LAMMB LABS: used each event to improve how the next one operated.

02 Follow the behavior

Actions are stronger evidence than opinions.

What people say matters. What they actually do usually matters more.

I try to separate stated problems from observed behavior and avoid treating every interview response as equally strong evidence.

Twenty-one interviews challenged the assumptions behind my original marketplace concept. Instead of defending the idea, I changed the product sequence.

See the research system behind this decision ↓

03 Prove it, then scale it

Scale what works, not what you hope works.

Scale amplifies whatever is already there. If the underlying model is wrong, more users, more money, or more complexity usually makes the problem larger.

I prefer finding the smallest credible way to test the important assumption first.

Nightlife: research before committing further to the marketplace model.

LAMMB LABS: individual events tested markets, acquisition channels, and economics before expansion.

Nêm Nếm: expanded capacity and pricing after demand was demonstrated through smaller sold-out dinners.

04 Own the outcome

Especially when the plan breaks.

Plans fail. Assumptions break. Systems behave differently in the real world than they did on paper.

I care less about whether the original plan survived than whether the problem got solved.

One hour before doors

The team lost access to the expected ticket-scanning system.

~1 hour

Replacement scanner built

3 phones

Running concurrently

~200

Check-ins

0

Recorded failures

The save mattered. What it revealed about the business mattered more.

Featured system

Research System

How do you stop AI-assisted research from becoming a collection of disconnected summaries?

I built a human-directed research system that keeps interviews, evidence, assumptions, contradictions, and product decisions connected as the research evolves.

21 interviewsEvidence hierarchyAssumption trackingAI-assisted synthesis
Close

The work changes shape. The pattern underneath is what connects it.

Back to work