Signals become structure

The preference layer for intelligent systems.Decoding subjective domains.

Palate Engine maps the judgments people make every day—what fits, resonates, belongs, and matters—so intelligent products can reason beyond relevance alone.

Join the team

Intelligence can generate nearly anything. The harder question is knowing what deserves to exist—and how to make it feel right.

A campaign with quiet confidence
Warm, not sweet
A product that feels inevitable

Mission

Human preference is not a single score.

It is a web of identity, context, memory, culture, and intention. Most recommendation systems flatten that richness into clicks.

We build models and agents that preserve the nuance—so software can understand not just what people choose, but why.

Make the unquantifiable useful without sanding it smooth.

Palate across the stack

Data
PE
Model

Preference models

Train models that understand the difference between fitting a pattern and making the right creative judgment.

research@company.comGet in touch
Which direction belongs?
Quiet editorialStrong fit
Bright utilityWeak fit
Warm modernismExplore

Preference agents

Give products and creative tools a reasoning layer for selection, synthesis, critique, and curation.

studio@company.comGet in touch

Read our research and field journal.

Essays, experiments, and technical work about modeling preference without reducing its complexity.

Stop averaging away the signal

Why consensus metrics erase the exact edge that makes preference useful.

Explore →

Preference is a moving target

A framework for context, drift, and decisions that change meaning over time.

Explore →

Selection is part of generation

The systems that create more need stronger judgment, not merely more output.

Explore →

Choose yes or no.

Preference becomes legible one considered choice at a time.

01
02
03
04
05
Keep itNot for me