The Method

We are not building memory.
We are building cognition.

Two sources of data. Together they capture not just what an operator decided, but how they reasoned.

01

What the operator deposits

The model learns directly from the operator. Recorded sessions, decision frameworks, scenario responses, written deposits.

First-person reasoning: how they work a problem, what they weigh, what they took from the calls that went wrong. The specific logic of how this particular mind decides.

02

What the people who worked alongside them observe

The people who watched the operator decide see what the operator would never think to record. A co-founder remembers how a deal was reasoned through. A long-tenured exec remembers the hire everyone else doubted.

The outside view captures patterns the operator takes for granted. In a family, this is a daughter noticing how her father handled pressure. Same mechanism.

03

The combination builds a cognitive model

Inside data plus outside data equals something neither could produce alone.

The archive becomes training data for the entity. The entity learns not just what they said but how they think. Not just their opinions but their reasoning patterns.

This is the distinction between memory and cognition. Memory stores facts. The entity captures how a specific mind works.

04

How the archive accumulates

Capture is ongoing, not a single interview. The operator works through 20 real business scenarios and a sequence of 29 decision questions across the 8 domains that built the company.

Each response adds depth. The longer the active period runs before a transition, the more the model has to work with. Begin before the handoff is on the calendar, not after.

The archive begins with The Founding, a 90-minute guided session with a Senior Legacy Guide.

What happens in The Founding

How We Evaluate What You Share

Not everything you deposit is equally valuable.

A specific story about a specific morning in 1974 is worth more than a general statement about hard work. Every deposit is evaluated to reflect that distinction.

First Review

Volume scoring

Speed and coverage

Every deposit

Second Review

Deep evaluation

Ambiguous cases

High-stakes decisions

Only after both systems agree does a deposit influence your entity

Your archive is not scored by the cheapest available model. It is scored by the right model for each decision.

06

Two layers, one handoff

The fingerprint is frozen at transition, fixed from the operator’s own deposits. It does not drift.

A living layer lets the successor add today’s context, while the judgment stays the operator’s.

See the full architecture

The second door

Where this began: families.

Basalith started as a way for families to preserve how someone thinks while they are still here to get it right. The method is identical. The operator becomes a parent. The colleagues become the family. The scenario library becomes everyday memory.

For family archives, an upload pipeline filters and ranks photographs before labeling, so the family only sees what is worth their time.

Begin a family archive