Automated rights-mapping workflows that save hundreds of policy research hours, reveal where protection doesn't exist, and direct funding to the states that have none.
Every state sets its own reproductive health rules, and attaches its own conditions to each. Wherever those rules differ, what a person can actually reach differs with them. Whether a right reaches them depends on the state, plan and the conditions attached, and no one has mapped that in full, leaving funders and advocates working from a partial picture leaving funders and advocates deciding where to work from a partial picture.
Inspired by Marli Ehrman
A right can exist and still not reach who is in need
A right can exist and still not reach who is in need
A right can exist and still not reach who is in need
Between a law being passed and a person using it sit several separate conditions. Every one of them has to hold and they have to navigate through this complex map to get those benefits.
Federal law may set a floor: For some rights it does. For others like ovarian cancer screening among them there is no federal rule at all, and whatever protection exists is fragmented by how individual states have chosen to write.
State may add to that floor, or narrow it: The same right can be guaranteed in one state, conditional in the next, and absent in the one after that.
The plan may reach, or sit outside: State insurance mandates do not touch self-funded employer plans, which is how most working people in the country are covered. A mandate can exist and simply not apply to the situation.
The eligibility conditions: Age, timing, prior authorization, documentation, consent each written separately within the law.
Every one of these has to hold true for the right to apply. It is very complex to navigate through these complexities without help, so the law is always written but may never reach those who need it.
Reproductive rights in the United States fall into a small number of broad categories. Four of them cover most of what a woman in this country actually faces: abortion, maternal health, assisted reproduction and
contraception. Each of those four breaks into narrower rights — eighteen in total, and that is the level the law is actually written at, and the level she runs into it.
Thirteen are covered by existing advocacy work. Five are covered by no one. All eighteen change by jurisdiction.
One Answer Takes Three Coordinates
Every rule we have found so far sits somewhere in the graph representation below. Between different kinds of right, four levels of jurisdictions. Dfferent ways of eligibility getting decided. One woman, asking one question, is a single dot in that space. Before anyone can tell her anything useful, they have to work out which dot that particular situation is.
Now take one of those dots and look closely.
Knowing her state gets you part of the way. So does knowing what she needs. So does knowing her situation. None of them answers her on its own. The answer sits where all three overlap.
And that overlap can be a lot smaller than it looks.
The strongest single asset in this landscape is the definitive
state-by-state map of abortion law. But abortion is one of eighteen subtopics, and it is the only one anyone publishes state by state. The other seventeen exist for a woman only as scattered statutes she cannot find.
We supply the layer that is missing. The 789 legal citations and the eligibility criteria behind them are already researched and already structured — enough to answer an individual woman in her own context, across all eighteen.
Which state is strongest for reproductive health?
Which state is strongest for reproductive health?
Which state is strongest for reproductive health?
We put that question, the easiest one to two large language models. Both named the familiar coastal cluster: California, Oregon, Washington, New York, Massachusetts.
The right answer is Illinois. Almost 89% coverage from a reproductive rights perspective is there with california being the second strongest.
When we asked why it missed it, one of them explained itself:
"A knowledge retrieval and structural bias failure, over-weighting the most frequently cited examples in general training data, while failing to systematically check for vital exceptions in other regions."
This is a big structural miss and states three things very clearly.
-It weights by how much has been written. The states with the most written coverage win, whatever the law actually says.
-It cites nothing you can check. No statute, no section, no effective date.
-It has no eligibility logic. It cannot tell her whether she personally qualifies.
If a general model misses the simplest question in this landscape, it is worth asking what it does with the harder ones.
ChatGPT and Gemini both named the familiar coastal cluster — California, Oregon, Washington, New York,
Massachusetts. Neither named Illinois, which is the answer. Asked why, one conceded a retrieval and structural bias failure: it repeats the most-cited examples instead of checking every region.
If a general model misses the simplest question here, it cannot be trusted with the harder ones.
Imagine what this unlocks.
Imagine what this unlocks.
Imagine what this unlocks.
Nothing like it exists today.
Nothing like it exists today.
Nothing like it exists today.
Inspired by Berthe Morisot
Inspired by Berthe Morisot