averagejoematt

a measured life · proof, not promises

An honest documentary of an ordinary life, rebuilt with AI.

One regular guy, the wearables he already owns, and an AI board reading every number he produces — published either way, down weeks shown, not hidden. The whole site is one feedback loop. Scroll it once and you know how everything here works.

day ·· measured in public

The measuring rule, bent into the loop the site runs on. Every station is a door you can walk through — and the cockpit is today's slice of all four at once.

station 01 · the data — reads

Every day writes its own numbers.

Sleep, training, glucose, weight, mood — pulled raw from the devices he already wears. Every source is named, and a gap shows as a gap, never a guess.

walk into the data →

station 02 · the coaching — proposes

An AI board reads them — and disagrees.

Coaches with different philosophies read the same day and argue about what it means, in public. Every call gets scored later, so the track records are real.

meet the coaching →

station 03 · the protocols — shifts

The arguments become experiments.

Supplements, protocols, challenges — each one a falsifiable bet that a specific change moves a specific number. The ones that fail stay on the page.

see the protocols →

station 04 · the story — narrates

And every week gets written down.

A chronicle, a podcast, his own journal — what the loop actually did, published on the good weeks and the bad ones alike.

read the story →

the loop closes · protocols shift the data

That's the whole site — and every number lands in public.

The ordinary life is Matthew's — every climb before this one ended the same way, and a drawer of devices measured everything and changed nothing. What's different this time is the loop above: every number read, argued over, and published either way.

This attempt starts at the Day‑1 weigh‑in, aimed at 185 lbs held for 90 consecutive days — or the checkpoint fails, on the record. The model tracks the odds, collapses included. The scale is an instrument, not the finish line.

No million-dollar lab. Just the wearables already on his body, and a model that reads the numbers back to him every morning.

Start with today in the cockpit → Follow the experiment

the pillars · measured co-movement

Seven areas of one life, pulling on each other.

Seven areas of life — he calls them pillars — each scored out of 100; node size tracks the score (small = low, a young experiment starts low, not broken). The lines show how two pillars actually move together in the measured data: a solid ember line means they rise and fall together, a dashed line means they trade off; thin or faint lines mean the pattern is weak, or there's only a few days of data behind it. A pillar with no lines hasn't shown a clear pattern with the others yet. Tap a pillar to open its evidence.

for family & friends · the short version

Is he okay this week?

A plain-language weekly status loads here. In the meantime, the live cockpit has the day's actual numbers.

See the full daily instrument →

the shape of it

The climb isn't a straight line. Here's every day, including the ones that dipped.

strong holding down — shown, not hidden no data tap any day → its cockpit

See today in the cockpit →

where it started · where it is

··lbs down
···lbs today
··%to goal

The full results & body composition →

try it on · your one number

Where would you land?

Type one of your own numbers and see where it sits next to Matthew's real band. Nothing you type leaves this page — no request, no cookie, no storage. And this is one subject's data (N=1), not a benchmark or health advice.

the third wall · ai says ↔ how it felt

The AI

Reading the latest field note…

Read the full Third Wall & meet the coaching team →

the story · the writing

The chronicle, the podcast, and the journal — in their own words.

    Read the full story — chronicle, podcast, journal, timeline & about →

    Meet the AI coaching team & their lab notes →

    the AI, made personal

    Ask the data anything.

    Type a real question and an AI answers from the actual numbers — in patterns, not diagnoses, rate-limited, honest when it's thin. Not a chatbot guessing; a model reading one real life's data back to you.

    The full Q&A page →

    An ordinary person did this with tools you already own.

    Not a guru. Not a lab. A model, some wearables, and the willingness to publish the bad weeks too.

    See what it's built on → Follow the experiment