Hormones are upstream of energy, sleep, mood, strength, focus, recovery and how well you age. They are also the one thing almost nobody measures. Baseline makes that measurable — and builds the evidence out of the people actually living it, rather than whoever is selling something.
The problem
People count steps and calories for years without ever looking at the signal that decides what the body does with either.
Energy, sleep, mood, recovery, fertility — these are not separate systems. They are your endocrine system expressing itself.
Most of what circulates about hormone health comes from someone with a product attached. Almost none of it carries numbers you can check.
Which is why so many people are doing everything correctly and still feel terrible — and get told there is nothing to look at.
Mission
Get hormones right and everything else gets easier. Get them wrong and nothing else fully works.
I believe optimising hormones is the highest-leverage thing a person can do for their health, because there is nothing downstream it does not touch.
Baseline exists to make that measurable, and to build the evidence out of the people actually living it. Not authority, not whoever is selling something — thousands of ordinary lives, recorded properly, over time.
That only means anything if it can go against me. If the data says something I believe does nothing, it goes in anyway. A record you cannot be wrong in front of is not worth keeping.
The urgency is generational. Our environment changed faster than our biology did, and whoever inherits it inherits what we did and did not bother to find out. That work starts now or it does not happen.
What it is
You log your own data. You review the products and practices you actually use. Over time, you join screened protocols. The dataset that emerges is the product — the social layer is how it gets collected.
A daily check-in that takes under a minute, plus your lab panels whenever you get them.
Your own trends against reference ranges. Two points are not a trend — twelve are.
A review says "my testosterone went up" and points at the measurements that prove it.
Evidence scores recompute from primary data — yours and the published literature, weighed separately.
Features
Baseline is being built in phases, deliberately. Everything in the first list is written and running in the pilot today.
One short check-in, with a streak. The whole product rests on whether this loop is worth returning to, so it is the only thing built so far.
Eleven hormone and metabolic markers, entered in whatever unit your lab printed. Date, time of day and fasting state come with it, because a cortisol reading without a time is not a reading.
Testosterone is ng/dL in the US and nmol/L nearly everywhere else — a factor of about 29. Values convert once, on the way in, and exactly what you typed is kept alongside forever.
A value far outside the reference range stops and asks "does this look right?" before it saves. That one dialog catches most unit mistakes at the point of entry, where they are still cheap.
Every marker charted against the range it should sit in, so a number means something the moment you look at it.
One nudge, at an hour you choose, that you can turn off in a tap. Not a notification stream.
Photograph your panel so you can read it while typing. The image stays on your device and is deleted the moment you save or cancel. There is no upload path, by design.
A complete CSV of your own data, on demand. Not a feature request — a starting condition.
A record of who accessed your data and why, running from the first day rather than added once there is something to explain.
Graded, cited answers to common hormone questions — with the actual numbers, and the study that disagrees. Open to anyone, no account needed.
The join that makes the whole thing work. A testimony references the observations that back it, so a claim of "doubled my T" can be checked against a panel that went 480 to 511 — automatically, by arithmetic, not by a moderator.
Supplements, light, cold, heat, sleep, clothing, food — one catalogue, each entry carrying an evidence score computed from the literature and from what members actually measured.
Every study sourced from PubMed with a verified DOI, summarised with its numbers rather than its vibe. Where two studies disagree, both are loaded and linked to each other. No winner is picked for you.
A hypothesis and a primary endpoint locked before anyone enrols, a fixed duration, and required disclosure of everything else that changed. Structured enough to mean something, run by people who volunteered.
What the population actually shows, in aggregate — including the parts that contradict the popular explanation.
Drop in the PDF, confirm the values it read, and it is logged. Plus Apple Health, Oura and Whoop for the daily signals that sit between panels.
Consented, exportable data for people doing real work — where consent is enforced by the database itself, and unconsented rows are not returned even if the application has a bug.
Licence and NPI checked against the public registries. "Dr." is blocked in a handle until it is verified, because impersonated authority is the most damaging thing that can happen to a community like this.
What you can track
Each one carries its reference range, its accepted units, and a note on how to make the reading actually comparable — morning draws, fasting state, cycle day.
Principles
These are enforced in the database, not promised in a policy page. Most of them cost something — that is roughly the point.
A measured value is never edited. Correcting one writes a new row and marks the old one superseded. A dataset where history can be quietly rewritten is not a research dataset, and researchers are right not to trust one.
Mixed units produce aggregates that look completely plausible and are completely wrong, and you do not notice for months. Keeping what you actually typed means a bad conversion is recoverable instead of permanent.
Who can see a row is decided by the database on every single query. Not by an application remembering to filter, and not by a company remembering its own privacy page.
Star ratings and evidence scores are computed separately and are never merged. A thing can be beloved and useless at the same time, and the interface has to be able to say so.
Publication bias is the failure mode most likely to quietly ruin a dataset like this. So the incentive is inverted at the source: a null result earns more standing than a positive one.
Typed by hand, read from a PDF, or delivered straight from the lab — each carries a different confidence, set by the path it arrived through. No setting anywhere lets a user raise it.
Being able to say that without flinching is the entire differentiator. Someone can feel dramatically better and be completely wrong about why, and a platform that cannot separate those two things is just another marketing channel.
Privacy & scope
This is health data about your body. The commitments below are structural — they were built in before there were any users to reassure.
Where it stands
Almost everything on the roadmap is more interesting to build than the logging loop. All of it waits anyway.
Accounts, the consent model, the data model, the brand. The security policies were written before the features they protect.
Daily logs, lab entry, unit conversion, trends, reminders, export. Live now as a private web pilot.
Fifty people, four weeks, and one number that decides what happens next.
Reviews linked to measurements, evidence scoring, the study library.
Protocols and enrolment, aggregate insights, PDF import, wearables.
The researcher console, consented exports, clinician verification.
The gate: 50 people, 4 weeks. At least 40% have to come back and log again in week two.
Above that, the next phase gets built. Below it, the loop itself is the problem and no amount of social features would rescue it. Nothing past the logging loop gets built until that number exists — because building on a guess is how you end up with a beautiful product nobody opens twice.
Get in touch
The pilot is small and deliberately so. If you measure your own hormones, or you want to start, say so below and you'll be first in when it opens.
Prefer email? finnrichert3@gmail.com