Human Experience Data Infrastructure
The data your models need
may not exist yet.
ProustMe originates permissioned human experiential data designed for authorized machine use.
Original • Structured • Provenance-Aware • Permission-Aware • Longitudinal
The public web is a record of what humans published.
It is not a record of everything humans know.
Many of the experiences most useful for understanding human reasoning, judgment, change and consequence were never published online.
They cannot simply be scraped.
They must first be expressed.
ProustMe creates new human data supply.
Conventional Data Pipeline
Existing information
↓
Collect
↓
Clean
↓
Label
↓
Transform
↓
Train / Evaluate
ProustMe
Human experience
↓
Elicit
↓
Originate
↓
Structure
↓
Establish provenance
↓
Apply permission
↓
Authorized machine use
One prepares data that exists.
The other originates data that doesn't.
Experience becomes structured.
ProustMe converts human-originated responses into structured experiential representations capable of preserving context around:
Events
People
Relationships
Decisions
Beliefs
Emotions
Temporal context
Outcomes
Change
Most datasets capture a moment.
ProustMe can capture what happens after it.
Longitudinal experiential data can preserve how human reasoning, circumstances and interpretation develop over time.
Static data describes a person.
Longitudinal data can describe change.
Authorized Applications
Built for machines that need to understand humans.
Model Evaluation
May support evaluating model behavior against structured human experiential context.
Alignment Research
May support research involving human values, preferences and lived outcomes.
Reasoning Research
Potential to study structured relationships among decisions, motivations and consequences.
Retrieval
May support retrieving relevant human experiential context.
Cultural Intelligence
Potential to represent variation across geography, language, generation and culture.
Personalization
May support permissioned human-context applications.
Agent Grounding
Potential to provide structured human context for systems acting on behalf of people.
Longitudinal Modeling
May support studying change across time rather than isolated observations.
Permission shouldn't live in a spreadsheet.
ProustMe is designed so provenance and permission can accompany human experiential data as part of the underlying architecture.
Origin
Who or what originated the information.
Provenance
How the information was captured and transformed.
Permission
What authorized uses apply.
Auditability
Records associated with authorized downstream use.
Responsible human data requires more than consent at collection.
It requires governable data throughout its lifecycle.
Synthetic data can scale computation.
It cannot create a childhood.
- A migration.
- A marriage.
- A failure.
- A diagnosis.
- A career.
- A loss.
- A decision.
- A reconciliation.
- A changed mind.
Human experience has a source.
ProustMe preserves that distinction.
Human intelligence is culturally situated.
173
Countries
11
Languages
ProustMe's global footprint creates the potential for experiential datasets representing different languages, cultures, generations and geographies.
The supply side is human.
ProustMe is building a contributor network around a simple principle:
People who originate the experiential asset should participate in the economics created from it.
50%
Human Experience Network
50%
ProustMe
This alignment is intended to support a durable source of high-quality human participation rather than one-time extraction.
The Origination Layer
The moat isn't access to the internet.
It's access to human experience that was never on it.
Patent Pending
Tell us what your models cannot learn from the internet.
Start with the human intelligence your system needs.
ProustMe can explore whether that need can be translated into an appropriate human-experience data objective.
The next great AI dataset
may not exist yet.
ProustMe exists to originate it.
PROUSTME
Human Experience Data Infrastructure.
We make models more human.