I am not a mom. I want to get that on the record before I say anything else, because what follows is a post about what working moms go through, and I am deeply aware that I am not one.
Over the past few years, I've watched pregnant people in my family, in my community, and in the course of this work try to navigate pregnancy alongside full-time jobs. Some had supportive employers, good benefits, flexibility, and strong family support. Some did not. What stood out was that even in relatively good situations, the logistics could be exhausting: prenatal appointments, paperwork, time off, benefit questions, decisions about when to tell work, and the constant effort of fitting all of it around a job.
And the consistent message from moms, especially those caring for infants or multiple kids, was that pregnancy is only one part of the story. What looks manageable on paper can become much harder once caregiving is part of everyday life. The gap between what employers say and what work actually feels like for caregivers is often wider than non-mom people realize.
This tool tries to help working moms, and anyone with caregiving responsibilities, figure out whether an employer is genuinely supportive of caregivers or just claims to be. And give actionable steps to improve.
I want to be clear that this tool is not going to change the world and it is not going to fix any of the actual hard stuff. It is a small, specific thing that might close a small, specific information gap. That's it.
But that got me thinking about something bigger. Software is full of assumptions. When those assumptions go unexamined, bias scales. So what if we brought subject-matter experts into the build itself, so that research-backed support scales instead of bias? Large tech companies do this — often for less than benevolent aims — but small companies and organizations rarely have the capacity to do anything but build and release as fast as they can.
How It Was Built
The process started with Google deep research into the current state of caregiver and maternal policy in this country. What federal and state protections exist, where the gaps are, what employers are required to offer versus what they volunteer, and what the data in peer reviewed journals and research says about where workers fall through the cracks (more about this later).
That research became the tool. The subject-matter expertise is embedded in the scoring weights, the question design, the category structure. Software that reflects evidence instead of assumptions. The research is the product. I took that research and my idea for the product to Claude Code planning mode in Visual Studio Code.
From there, the plan went through engineering review. Experienced engineers read it, asked hard questions, and caught things that would have been missed otherwise. The data schema had real gaps. Some assessment logic would have made the tool a pain to maintain. The version you see now is about the fourth or fifth real iteration. The tool will get meaningfully better once actual working moms get their hands on it and tell us what's missing.
You can try it here: can-a-mom-work-here.labs.wherewego.org
Why This Matters
Software and AI systems lock in assumptions at scale. When those assumptions are wrong about who is competent, who is committed, who deserves flexibility, bias scales with them. Hiring algorithms that penalize career gaps. Performance systems that reward face time. Benefits platforms that bury the policies people need most. Employers that say they support moms but don't have a single mom in a leadership position.
This tool tries to go the other direction. It encodes research. CDC occupational health guidelines, maternal retention data, the documented mechanics of the motherhood penalty. The subject-matter expert isn't advising the software from the outside. The expertise is the software.
The scope is deliberately small. A scorecard. Sixteen questions. Deterministic scoring you can see through. But it's testing a real thing — does the quality of the knowledge baked into software determine whether it helps or harms? And if that knowledge can be delivered in small, actionable chunks, like a score, a recommendation, a specific thing to ask your employer, then maybe it actually reaches the people who need it.
The Research Behind the Score
The assessment evaluates workplaces across 7 research-backed dimensions of maternal support. Each dimension is weighted based on its documented impact on maternal workforce retention and wellbeing.
These weights are informed by the research framework I created with deep research called "Determinants of Maternal Occupational Sustainability," a synthesis of current evidence from CDC/NIOSH occupational health guidelines, the Pregnant Workers Fairness Act (PWFA), the PUMP Act for nursing mothers, SHRM employee benefits data, and maternal retention research from Maven Clinic, McKinsey, and academic studies published in peer-reviewed journals.
What I Take From This
Two things stood out in this process. Tools built with AI are only as good as the understanding you give them. The time I spent sitting with that research made everything downstream dramatically better. Second, the engineering review is not optional. The reason we can move quickly is that we have experienced engineers backstopping the whole thing. Take that out and you get fragile software that breaks the first time a real user does something you didn't anticipate.
What you get with both pieces in place is a useful tool built in days instead of months. And in this case, a tiny bit of information infrastructure that might save a mom somewhere a bad decision about her next job or help improve her current one. That's the whole thing. Happy Mother's Day.
You can see the tools as they launch, and tell us what we should build next, at labs.wherewego.org.
Shout out Gail Ifshin and Rebekah Mills-Ifshin — Happy Mothers day!
