Writing from insidethe behavioral engine
Writing from inside the behavioral engineWorking notes rather than marketing posts. Each entry documents a decision, a measurement, or a failure that shaped how the Vision Suite behaves.
Six notes on measurement, initiation and repair.
Scroll the rail for the newest writing. Every note is drawn from logged cohort data rather than general wellness advice.
Time blindness is a measurement problem, not a discipline problem
Most advice for time blindness assumes the person can already feel duration passing and is simply choosing to ignore it. That assump…
Executive function · Aug 11, 2026 · 9 min read
Repair language fails when it arrives without context
Scripted apologies collapse because they are written for a generic dynamic. Drawing on logged ruptures across our cohort, we examine…
Relational load · Jul 29, 2026 · 11 min read
Contraction is data, not failure
Streak mechanics teach people that a reduced day is a broken day. That framing is the single most common reason recovery routines ar…
Recovery pacing · Jul 15, 2026 · 8 min read
Why the assessment layer ships before the interface
We describe the internal rule that no recommendation surface may launch before its measurement layer is complete, including the two…
Product engineering · Jun 30, 2026 · 12 min read
What we refuse to do with behavioral data
A plain accounting of the commitments behind the privacy page: no identifiable training of public models, no advertising use, perman…
Privacy · Jun 12, 2026 · 7 min read
The first physical action is the only planning step that matters
Task lists fail at the initiation boundary, not at the organization boundary. We break down how the friction router selects a single…
Executive function · May 27, 2026 · 10 min read
Every note in full.
Each entry states the decision it documents, the measurement behind it, and what changed in the product as a result.
Executive function
Time blindness is a measurement problem, not a discipline problem
Most advice for time blindness assumes the person can already feel duration passing and is simply choosing to ignore it. That assumption produces useless interventions. We walk through how the engine converts subjective duration into logged intervals, why a two-week capture window is the minimum before any pattern is trustworthy, and how the resulting data changes the shape of a workable day.
August 11, 2026 · 9 min read
Relational load
Repair language fails when it arrives without context
Scripted apologies collapse because they are written for a generic dynamic. Drawing on logged ruptures across our cohort, we examine why the same phrase de-escalates one partnership and inflames another, what the escalation-speed variable actually predicts, and how to write a repair that references a specific event rather than an abstract feeling.
July 29, 2026 · 11 min read
Recovery pacing
Contraction is data, not failure
Streak mechanics teach people that a reduced day is a broken day. That framing is the single most common reason recovery routines are abandoned in month two. This piece explains how the pacing model records a contracted day as a legitimate input, what the trend line looks like when contractions are respected, and why the long curve improves faster without enforced consistency.
July 15, 2026 · 8 min read
Product engineering
Why the assessment layer ships before the interface
We describe the internal rule that no recommendation surface may launch before its measurement layer is complete, including the two features we cut for violating it. The piece covers the structure of the baseline instrument, how variables are versioned so historical data stays comparable, and the retrieval window the companion uses when answering a question.
June 30, 2026 · 12 min read
Privacy
What we refuse to do with behavioral data
A plain accounting of the commitments behind the privacy page: no identifiable training of public models, no advertising use, permanent deletion on request, and a processor list we publish rather than bury. We also cover the commercially attractive options we declined and the reasoning that made those decisions straightforward.
June 12, 2026 · 7 min read
Executive function
The first physical action is the only planning step that matters
Task lists fail at the initiation boundary, not at the organization boundary. We break down how the friction router selects a single physical action from a monolithic task, the capacity threshold at which expanding the ladder becomes counterproductive, and what happened when we tested a version that showed the whole sequence at once.
May 27, 2026 · 10 min read
Begin Your Transformation.
Reading about the method is useful. Running it against your own baseline is where the change shows up.