AI apps for women: which ones help, and with what?

Most AI apps marketed to women do one of six jobs. They track a cycle or fertility, support mental wellbeing, map safer routes, connect you with other women, organise family logistics, or help with what you write to people. Which one helps depends on the job. This piece sorts eleven apps by the job they do, says what stands behind each one’s claims, and lists what to check before you give any of them your data.

I co-founded Subtext, an AI writing assistant, with Anton Licht. It is one of the apps below, so weigh what I say about it accordingly. Everything else comes from each app’s own store listing or policy, from regulator records, from studies or their abstracts, and from press releases and news reports, as they read on 13 October 2026. The source list says which. What didn’t hold up is listed under Numbers I left out.

What counts as an AI app for women?

By an AI app for women I mean an app that runs on some kind of machine learning and is either marketed to women or does a job that, in the research further down, falls mostly to women. The kind of AI varies a lot. Natural Cycles runs an algorithm on your temperature¹. Wysa is a chatbot². Ohai reads school emails and pulls out the dates³. Subtext uses a language model to estimate how a message you wrote is likely to land.

So a star rating on one tells you little about another, and no rating tells you whether an app works. Regulation is a clearer signal. Two of the apps here have an FDA decision behind a birth-control feature. Natural Cycles is sold as FDA-cleared⁴, and Clue’s maker holds a 2021 clearance for a feature called Clue Birth Control, which I could not find on its US listing on 13 October 2026⁵⁶.

Most of these apps are also built around a stage of life, such as trying to conceive, being pregnant or becoming a parent. Communication runs through every stage.

Why does mainstream AI keep missing women’s work?

On 5 October 2026 (US Pacific time) Sonia Baschez, whom Business Insider describes as a marketing entrepreneur and podcaster, asked on X who is building an AI that carries the mental load instead of waiting to be given a chore⁷⁸. In my paraphrase, she wants an assistant that sees a birthday party coming before anyone has bought a present, reorders what is running out and books the dentist. What she keeps getting are agents that book a table when told to⁷. The newsletter Ben’s Bites embedded the post, which showed about 791,000 views there⁹, and Business Insider ran a column on it⁸. In can AI reduce the mental load? I looked at whether any assistant can do what she asked, and as of 12 October 2026 I found none that has shown it.

Part of the reason may be who decides what gets built, and who pays for it.

Who builds it. LinkedIn found that women made up 26% of US AI hires in 2025, against 50% of hires into other jobs. Across 27 countries, women hold 13% of C-suite AI roles at AI companies¹⁰.

Who funds it. US startups with at least one female founder raised a record $73.6 billion in 2025, according to PitchBook¹¹. More than $30 billion of that went to two companies, Anthropic and Scale AI. Fortune wrote, “Remove those two names, and the record disappears”¹². Teams made up only of women raised about $3.2 billion, 22% less than the year before¹³. (Disclosure: Subtext runs on Anthropic’s models.)

Who uses it, and what it costs them. A Harvard Business School working paper pooled 76 sources with 318,924 respondents in over 100 countries. It puts the use of generative AI at 47.8% for men and 39.3% for women. The authors call that a relative gap of 22% across the whole pool and of roughly 16% since early 2025¹⁴. Among ChatGPT’s own users the split has evened out. OpenAI says that among ChatGPT users whose first names could be classified, the share with typically feminine names rose from 37% in January 2024 to 52% by July 2025¹⁵. At work, the advocacy group Lean In found differences in how AI use is received. In its survey of 1,015 US adults, published in May 2026, 29% of women worried about being seen as cheating when they use AI, against 22% of men. Among people who had used AI at work, men were more often praised for it (23% against 18%), and men were more often encouraged by their managers to use it (37% against 30%)¹⁶.

Whose view shapes the design. In a Pew survey of AI experts run in late 2024 and published in April 2025, 75% said the people designing AI take men’s perspectives into account at least somewhat well. Only 44% said the same about women’s¹⁷.

Meanwhile, occupations where women are the majority sit closer to the tools. The ILO’s March 2026 brief found 29% of female-dominated occupations exposed to generative AI, against 16% of male-dominated ones¹⁸. Female-dominated occupations are more exposed to AI, and fewer women decide what it does.

The AI apps for women, sorted by the job they do

The table is the short version. The details and the catches follow it.

Cycle and fertility. Flo shows 4.7 stars from about two million ratings on the US App Store. It predicts periods and ovulation. Its Premium tier adds what Flo calls a “24/7 virtual Health Assistant” and a symptom checker with information on PCOS (which the listing calls by its newer name, polyendocrine metabolic ovarian syndrome), endometriosis and fibroids. The listing describes the checker as educational information, so it is no diagnosis, and it isn’t available in the UK, the EU or four other European countries¹⁹. Flo’s Anonymous Mode is built so that no single party sees both who you are and your health data, and Flo says it sped up the launch after the US Supreme Court overturned Roe v. Wade²⁰. Flo has a history with regulators on this. In 2021 the US Federal Trade Commission finalised an order over Flo sharing health data with marketing and analytics firms, including Facebook and Google²¹. In August 2025 a California jury found that Meta had broken California privacy law by collecting menstrual health data from Flo users, and Flo itself settled²².

Clue, which is based in Berlin, predicts periods, symptoms and fertile windows⁶. It says all of its data is stored on servers in the EU and asks for explicit consent before it handles health data. Its policy also allows transfers to countries outside the European Economic Area²³. Clue’s maker states on the US listing that the app is a regulated medical device⁶. Two FDA records sit behind that, and they differ. In February 2021 the FDA cleared a contraception feature called Clue Birth Control⁵. I could not find that feature on the US listing on 13 October 2026, where the birth-control section is about logging the method you already use⁶. The entry for the Conceive mode in the FDA’s device identifier database says it is exempt from premarket submission, which means the FDA does not ask for a review before it goes on sale. So “regulated” is accurate, and as I read the record the FDA has not assessed how well Conceive predicts ovulation²⁴. A predicted fertile window is not birth control unless the app is cleared for that use. Clue Plus costs $9.99 a month or $39.99 a year in the US⁶.

Natural Cycles is sold as birth control. In 2018 the FDA allowed it to be marketed as the first direct-to-consumer app for contraception¹, and in July 2026 it cleared an updated version that uses a machine-learning model and can add new wearables²⁵. Cleared means the FDA let it be sold for that use. It is no promise that the method suits you. In the studies the FDA reviewed, 15,570 women used the app for an average of eight months. With perfect use, 1.8 in 100 would become pregnant in a year, and with typical use 6.5 in 100. The FDA adds that no form of contraception works perfectly, that on the days the app marks as fertile you have to abstain or use protection, and that it does not protect against sexually transmitted infections¹. Talk to a doctor or nurse before you rely on any app for contraception. The app takes your temperature from its own band, a basal thermometer, an Oura ring, WHOOP, some Garmin watches or an Apple Watch Series 8 or newer. It costs $149.99 a year or $21.99 a month in the US⁴.

Mental wellbeing. Wysa is a chatbot that walks you through exercises from cognitive behavioural therapy (CBT) and dialectical behaviour therapy (DBT), with paid sessions with a human coach on top². A 2026 trial by Meinert and colleagues randomised 76 people referred to NHS Talking Therapies and found no evidence that Wysa treats depression. The authors add that the small sample could have influenced that result, and 30 of the 76 were lost to follow-up, so the trial cannot show the opposite either²⁶. A release note on the app’s listing dated 31 August 2026 says Wysa “is intended for wellness purposes”². Soula calls itself an AI companion made for women, with check-ins you can do in a few taps and paid modules for trying to conceive and pregnancy. Asked whether it’s therapy, its listing answers no²⁷. Either may help on a rough evening. Neither replaces care. If low mood or anxiety lasts, see a doctor, and in a crisis call your local emergency number or a crisis line.

Safety and community. My Safetipin was started in India in 2013 by Kalpana Viswanath and Ashish Basu²⁸. It scores streets on things like lighting, openness and public transport, and suggests safer routes. It’s free. It relies on people auditing the places you walk, so it works best where many people have done so, and its US listing shows 11 ratings²⁹. Peanut is a social app for meeting other mothers at the same stage. Its US listing speaks of “all stages of motherhood” and says every profile is checked with selfie verification³⁰. The core app is free, and its value comes from the people on it more than from any AI.

Family logistics. This is the job Baschez asked about, and two products in it have already ended or announced their end. Ohai reads school emails and turns them into calendar events and tasks, from $9.99 a month³. Cozi’s Max plan, at $79 a year, adds AI event import, a recipe creator and a meal planner³¹. Meta launched Muse, an agent that fills in forms and runs errands for you, in the US on 8 September 2026³². In early reviews it recommended places that had closed years earlier, and in one test it made up a phone number³³³⁴. Milo’s founder said goodbye to the product on 12 January 2026, writing that the technology is “simply too early to be reliable in any useful way”³⁵. Maple says it has been acquired and will shut down on 31 December 2026³⁶. Before you put a family calendar into any of these, check how you’d get it back out.

The part of the load made of words

A lot of the mental load ends in a message. Remembering the party means texting the other parent about a gift. Spotting that a dentist visit is overdue means rebooking it. Most of the apps above stop just before that step.

Who carries it. The sociologist Allison Daminger interviewed the members of 35 couples and split household thinking into four steps: anticipating a need, finding options, deciding, and checking it got done. In her sample, women did more of the anticipating and the checking³⁷. In a 2023 survey of 3,000 US parents, Ana Catalano Weeks at the University of Bath and Leah Ruppanner at the University of Melbourne asked who mostly handles each of 21 mental-load tasks. Mothers answered “mostly me” for 71% of them and fathers for 45%. For daily tasks it was 79% and 37%³⁸³⁹. The respondents came from different households and some were single parents, so the two figures were never meant to add up to 100. Keeping the wider family in touch appears to fall the same way. Researchers have called that work kinkeeping since the 1980s, and a 1993 study by Gerstel and Gallagher found it was done mostly by women, though men did some⁴⁰. I go through that research in why family messages feel like a second job.

What it costs at work. Kieran Snyder read 248 performance reviews from tech companies. Criticism of personality, with words like “abrasive”, showed up in 71 of 94 critical reviews of women and 2 of 83 critical reviews of men⁴¹. It’s a volunteered sample and wasn’t peer reviewed, but it fits Catalyst’s 2007 research on the double bind, where women leaders were seen as competent or likeable and rarely both⁴². In studies by Linda Babcock and colleagues, women were asked more often to take on work that doesn’t lead to promotion, and said yes more often. When 3,271 faculty at a large US public university were emailed a request to volunteer for a faculty senate committee, 7.0% of the women volunteered, against 2.6% of the men⁴³. So “I can’t take this on” is a message many women have to send, and the studies above suggest its wording carries more risk for them.

What text does to tone. In email experiments by Justin Kruger, Nicholas Epley and colleagues, people writing sarcastic or serious messages overestimated how well their tone would come across⁴⁴. Handing the words to someone else has its own cost. In a 2023 Ohio State experiment with 208 adults, people who imagined getting an AI-assisted message from a friend rated the friend’s effort lower, were less satisfied with the friendship and felt less sure where they stood. Help from another person had much the same effect⁴⁵. “Effort is very important in a relationship,” said lead author Bingjie Liu⁴⁶. That finding covers any help with the words, Subtext’s included.

Where Subtext fits, and when you shouldn’t use it

Subtext is for the one message you keep rewriting. You give it a draft, a screenshot of the thread or a voice note. It names up to three issues in your draft, marks the words at fault and names the feeling underneath⁴⁷. When a message really matters or the goal is unclear, it asks what you want to get across before it writes anything. Then it offers up to three versions, each flagged safe to send or check before sending. Behind the flag is a score from 5 to 100, which is Subtext’s own estimate of how likely a version is to land as meant. Each version shows how much of your wording it changed, so you can see which stays closest to your own words. Warmer, Shorter, More casual, Add humor and Add confidence are one tap each. Try Subtext in your browserDownload on the App StoreGet it on Google PlayGet the app

A test draft shows how. On “Sorry to bother you, but could you maybe take a quick look if you have time? No worries if not!” Subtext returned two tags. The first was Self-deprecation, with the note that “Sorry to bother you” “puts your request down before it’s even made”. The second was Clarity, with the note that the draft “doesn’t say what you’d like them to look at, so they can’t tell how big the ask is”. It named the feeling as Hesitance. The first of its three versions read “Could you take a quick look when you have a moment? No pressure if it’s not a good time.” That is the app’s real output, unedited. The version still doesn’t say what to look at. Subtext can’t know that, so you add it.

The score is Subtext’s own estimate. It can’t tell you how a particular person will react, and our research page says that “No study on this page tested Subtext”⁴⁸. Nobody independent has evaluated it.

Why it’s on this list. Subtext does a job the other ten don’t. Clue tracks your cycle, and the message to your manager about working from home on a bad day is still yours to write. Peanut finds you other mums, and the reply to the one who cancels every time is yours to write too. Subtext is built for anyone. The research above suggests that women do more of the work these messages come from and are criticised more for tone, which is why it belongs here. None of that is a fault in how women write. The research is about how they are read, and a wording tool is a workaround.

When not to use it. Subtext doesn’t keep track of the birthday, watch the calendar, buy the gift or send anything. It runs as its own app with no keyboard extension, so it only reads what you hand it, and you copy the version you want and send it yourself. It is a communication assistant, not therapy, and it gives no medical advice. After a few free analyses it needs a paid subscription. If the other person would mind that an app helped, write the message yourself. Check that your employer allows it before you paste work messages, and remember that a screenshot carries someone else’s words. Our terms leave both to you⁴⁹. A general chatbot will also redraft a message, and if your problem is a high volume of routine replies it will be quicker, as I explain in the best AI reply app depends on whether your problem is volume or stakes.

Privacy, checked against our policy on 13 October 2026. Nothing you send to Subtext is used to train AI models. Conversations are deleted five days after you last use them, or 90 days if you pin them, and deleting one yourself removes it from our servers straight away. Anthropic processes text, images and voice transcripts, Deepgram turns voice recordings into text, and OpenAI generates read-aloud audio if you switch that on. Conversations are stored on Google Firebase in the United States. Mixpanel receives product events tied to your user ID, our attribution provider Adjust receives device and campaign identifiers and may pass measurement data to ad networks, and RevenueCat receives your user ID, email and subscription status. With web search switched on, search terms drawn from your request go through Anthropic to search providers. We don’t sell your data and we show no ads in the app. We don’t have a zero-retention arrangement with any of the AI providers, so their own deletion windows apply. Anthropic, for example, deletes inputs and outputs within 30 days, though it may keep a request its safety systems flag for up to two years⁴⁹.

What to check before you give an app your data

Six questions cover most of it.

• Does it sell or share what you tell it? Flo’s 2021 FTC order and the FTC’s 2023 case against the ovulation app Premom, which ended with a proposed $100,000 civil penalty and a ban on sharing health data for advertising, both turned on sharing²¹⁵⁰. Look for a promise not to sell data and a list of who processes what.
• Where does your data live? Clue says it stores its data on EU servers²³. Subtext stores conversations in the US⁴⁹. Either can be fine, as long as you know which.
• Does it train on what you write? Policies differ. I compared writing tools in which AI writing assistants train on your messages?
• What is the claim backed by? “FDA-cleared” (Natural Cycles), “a regulated medical device” (Clue), “intended for wellness purposes” (Wysa) and not therapy (Soula, Subtext) promise very different things. Ask which feature a clearance covers.
• Does it act for you, or with you? An agent that books and buys needs more checking than a tool that drafts and leaves the sending to you. The seven questions in how can you tell if an AI assistant is proactive? help here.
• What happens if it closes? Milo ended in January 2026 and Maple closes on 31 December 2026³⁵³⁶. Check there’s an export before you commit.

Numbers I left out

These figures and claims turned up in write-ups about women and AI while I researched this piece. I couldn’t confirm them, or the source says something different.

• The size of the femtech market. For 2025 alone, market-research pages range from $45.6 billion to $66.2 billion, and none shows the data behind its figure.
• “29% of women’s jobs are exposed to AI.” The ILO figure is 29% of female-dominated occupations, from a March 2026 brief¹⁸. It appears above with that wording.
• “42% of people hesitate to use health apps over privacy.” I could only trace it to a market-report page with no method.
• “Only 31% of 521 young women have used a mental health app.” I couldn’t find the study.
• “Women have 22% lower odds of using generative AI.” That wording is from the 2024 version of the Harvard working paper. The 2026 update, with more than twice as many people, reports a relative gap of roughly 16% since early 2025, so the single older figure overstates today’s gap¹⁴.
• “Clue Conceive is FDA-cleared.” Clue’s 2021 clearance is for a different feature, Clue Birth Control⁵. The entry for Conceive in the FDA’s device identifier database says it is exempt from premarket submission²⁴.
• “Built-in phone writing tools work like keyloggers.” Nothing I found supports it.

Questions people ask

What is the best AI app for women? It depends on the job. For birth control, Natural Cycles is sold as FDA-cleared⁴. For cycle tracking, Clue says it stores data on EU servers, though its policy allows transfers outside the European Economic Area²³. Wysa is a chatbot with exercises and Soula offers quick check-ins, My Safetipin maps safer routes where it has data, Ohai and Cozi help with school and family admin, and Subtext, which I co-founded, helps with the message you can’t get right.

Are period tracking apps safe to use? On privacy it varies, so read what they share. Look for a promise not to sell data, an anonymous mode or EU storage, and a clean regulatory record. Flo’s 2021 FTC order and Premom’s 2023 case show what goes wrong when sharing isn’t disclosed²¹⁵⁰. On reliability, a predicted fertile window is not birth control unless the app is cleared for that use.

Is there an AI app that carries the mental load? Not yet, as far as I can tell. Family assistants can read school emails and fill a calendar, but noticing what’s needed and following up still sit with you, and of the family assistants I looked at, one ended in January 2026 and another closes on 31 December 2026³⁵³⁶. I go through what the big general assistants can do in can AI reduce the mental load?

Can AI help me say no or set a boundary? It can help with the wording. Subtext marks the words that could land wrong, and Add confidence gives you a firmer version in one tap. You decide what to send. For the research on whether boundaries help, see is setting boundaries actually good for you? and how do you say no to extra work?

Is Subtext only for women? No. Subtext is for anyone with a message that matters. It’s in this piece because the research above suggests women do more of the work those messages come from and are criticised more for tone.

Sources

Numbered in order of first appearance. Pages were checked on 13 October 2026. Store ratings and prices are from the US App Store on that date and change often.

1. FDA allows marketing of first direct-to-consumer app for contraceptive use to prevent pregnancy. US Food and Drug Administration. 10 August 2018. . Regulator press release on a De Novo marketing authorisation; source of the failure rates; read in an archived copy of 31 December 2024 because fda.gov blocks automated readers. Checked 13 October 2026.
2. Wysa: Mental Wellbeing AI. US App Store listing. . Vendor listing; the wellness wording is in the version history, release note of 31 August 2026. Checked 13 October 2026.
3. How it works. Ohai. https://www.ohai.ai/how-it-works/. Vendor page. Checked 13 October 2026.
4. Natural Cycles. US App Store listing. . Vendor listing; supported devices and prices as shown on the check date. Checked 13 October 2026.
5. 510(k) premarket notification K193330, Clue Birth Control. US Food and Drug Administration. Decision dated 18 February 2021. . Regulator database; applicant Biowink GmbH; decision “substantially equivalent”. Checked 13 October 2026.
6. Clue Period & Cycle Tracker. US App Store listing. . Vendor listing; features, medical-device statement and prices as shown on the check date. Checked 13 October 2026.
7. Post on X asking who is building an AI that carries the mental load. Sonia Baschez, X. 6 October 2026, 00:30 UTC (5 October, US Pacific time). . Social post; the view count is from the Ben’s Bites embed. Checked 13 October 2026.
8. I don’t need an AI agent to be an executive assistant. I need it to be a wife. Katie Notopoulos, Business Insider (syndicated by Yahoo Style UK). 9 October 2026. . Read in the syndicated copy. Checked 13 October 2026.
9. Text is so 2023. Ben Tossell, Ben’s Bites. 8 October 2026. . Newsletter; embeds Sonia Baschez’s post on X, showing about 791,000 views at an unknown time. Checked 13 October 2026.
10. New LinkedIn research finds women account for just 26 percent of AI hires as AI jobs surge. LinkedIn News. 18 August 2026. . Company research on its own platform data. Checked 13 October 2026.
11. AI Megadeals and Capital Concentration Define 2025 for Female Founders Across US and Europe. PitchBook, via Business Wire. 5 March 2026. . Press release on PitchBook’s All In reports; read in a syndicated copy because the Business Wire page blocks automated readers. Checked 13 October 2026.
12. Venture dollars to female founders doubled to a record $73 billion last year, but Anthropic and Scale AI skewed the data. Lily Mae Lazarus, Fortune. 6 March 2026. . News report; source of the quote. Checked 13 October 2026.
13. Female funding sets record nationally, but there’s more to this story. WLRN. 9 March 2026. . News report; secondary source for the all-female-team figure. Checked 13 October 2026.
14. The gender divide in generative AI: a global challenge (update). Cranney, Delecourt and Koning, Harvard Business School AI Institute (formerly the Digital Data Design Institute at Harvard). 6 August 2026. . Summary of HBS Working Paper 25-023 (SSRN 6880085); pools 76 sources; not peer reviewed. Checked 13 October 2026.
15. How people are using ChatGPT. OpenAI. 15 September 2025. . Vendor research on its own users; gender inferred from first names that could be classified. Checked 13 October 2026.
16. Survey findings on women, AI and the gender gap. Lean In, with Wired Research. Published 7 May 2026. . Survey of 1,015 US adults fielded 2 to 6 March; advocacy organisation. Checked 13 October 2026.
17. How the US public and AI experts view artificial intelligence. Pew Research Center. 3 April 2025. . Survey report; 1,013 AI experts surveyed 14 August to 31 October 2024. Checked 13 October 2026.
18. New ILO data confirm women face higher workplace risks from generative AI than men. International Labour Organization. 5 March 2026. . Press release on an ILO brief. Checked 13 October 2026.
19. Flo Cycle & Period Tracker. US App Store listing. . Vendor listing; rating, features and regional limits as shown on the check date. Checked 13 October 2026.
20. Anonymous Mode. Flo Health. . Vendor page. Checked 13 October 2026.
21. FTC finalizes order with Flo Health, a fertility-tracking app that shared sensitive health data with Facebook, Google, and others. US Federal Trade Commission. 22 June 2021. . Regulator press release. Checked 13 October 2026.
22. Jury rules Meta violated California privacy laws by collecting menstrual health data from Flo. TechCrunch. 5 August 2025. . News report; also reports Flo’s settlement. Checked 13 October 2026.
23. Clue privacy policy. Clue (BioWink GmbH). Updated 29 September 2026. . Vendor policy. Checked 13 October 2026.
24. Device record for Clue Conceive. US FDA Global Unique Device Identification Database (AccessGUDID). . Regulator database; entries are supplied by the device’s labeller. The record reads “Device Exempt from Premarket Submission: Yes”. Checked 13 October 2026.
25. 510(k) premarket notification K253694, Natural Cycles. US Food and Drug Administration. Decision dated 17 July 2026. . Regulator database; decision “substantially equivalent”. Checked 13 October 2026.
26. Real-world testing of an AI conversational agent as an early intervention and support tool in the mental health referral care pathway: a mixed-methods study. E. Meinert, M. Milne-Ives, E. Taylor and colleagues. International Journal of Social Psychiatry 72(6): 1819 to 1832, 2026. . Peer reviewed; randomised controlled trial, 76 people randomised 2 to 1; read in the abstract. Checked 13 October 2026.
27. Soula: Women’s Mental Health. US App Store listing. . Vendor listing. Checked 13 October 2026.
28. Safetipin. Wikipedia. . Secondary source for the founding details. Checked 13 October 2026.
29. My Safetipin. US App Store listing. . Vendor listing. Checked 13 October 2026.
30. Peanut: Find Mom Friends. US App Store listing. . Vendor listing. Checked 13 October 2026.
31. Compare plans. Cozi. https://www.cozi.com/compare-plans/. Vendor page. Checked 13 October 2026.
32. Introducing Muse: The World’s First Personal AI Agent Built for Everyone. Meta Newsroom. 8 September 2026, modified 30 September 2026. . Vendor announcement; “World’s First” in the title is a marketing claim. Checked 13 October 2026.
33. Meta’s Muse is an impressively capable AI agent, despite some hiccups. Daniel Howley, Yahoo Finance. 23 September 2026. . Product review. Checked 13 October 2026.
34. Meta says its Muse AI agent can do things for you. I put it to the test. Lisa Eadicicco, CNN. 23 September 2026. . Product review. Checked 13 October 2026.
35. hello/goodbye. Avni Patel Thompson, Milo (Substack). 12 January 2026. . Founder’s announcement. Checked 13 October 2026.
36. Maple homepage, closure notice. Maple. . Vendor page; says Maple has been acquired by Wander and will sunset on 31 December 2026. Checked 13 October 2026.
37. The Cognitive Dimension of Household Labor. Allison Daminger, American Sociological Review 84(4): 609 to 633. Online 9 July 2019. . Peer reviewed; qualitative, 70 interviews with members of 35 couples. Checked 13 October 2026.
38. A typology of US parents’ mental loads: Core and episodic cognitive labor. Ana Catalano Weeks and Leah Ruppanner, Journal of Marriage and Family 87(3): 966 to 989. Online 12 December 2024. . Peer reviewed; quota survey of 3,000 US parents fielded in 2023, including single parents. Checked 13 October 2026.
39. Mothers bear the brunt of the mental load, managing 7 in 10 household tasks. University of Bath. 12 December 2024. . University press release on the study above. Checked 13 October 2026.
40. Kinkeeping and distress: gender, recipients of care, and work-family conflict. N. Gerstel and S. K. Gallagher, Journal of Marriage and the Family 55(3): 598 to 607. 1993. . Peer reviewed; read via the AGRIS abstract record; sample not stated in the record; correlational. Checked 13 October 2026.
41. The abrasiveness trap: high-achieving men and women are described differently in reviews. Kieran Snyder, Fortune. 26 August 2014. . Magazine article on a volunteered sample of 248 reviews; not peer reviewed. Checked 13 October 2026.
42. The double-bind dilemma for women in leadership: damned if you do, doomed if you don’t. Catalyst. 2007. . Research report by an advocacy organisation. Checked 13 October 2026.
43. Gender Differences in Accepting and Receiving Requests for Tasks with Low Promotability. Linda Babcock, Maria P. Recalde, Lise Vesterlund and Laurie Weingart, American Economic Review 107(3): 714 to 747. 2017. . Peer reviewed; field data on 3,271 faculty. Checked 13 October 2026.
44. Egocentrism over e-mail: can we communicate as well as we think? Justin Kruger, Nicholas Epley, Jason Parker and Zhi-Wen Ng. Journal of Personality and Social Psychology 89(6): 925 to 936, 2005. . Peer reviewed. Checked 13 October 2026.
45. Artificial intelligence and perceived effort in relationship maintenance: Effects on relationship satisfaction and uncertainty. Bingjie Liu, Jin Kang and Lewen Wei, Journal of Social and Personal Relationships 41(5): 1232 to 1252. Online 18 July 2023. . Peer reviewed; online experiment, 208 participants, hypothetical scenarios with a fictional close friend. Checked 13 October 2026.
46. AI can help write a message to a friend, but don’t do it. Ohio State News. 11 September 2023. . University press release; source of the quote. Checked 13 October 2026.
47. Subtext. Product page. . Our own claims. Checked 13 October 2026.
48. Research behind Subtext. . Our own page. Checked 13 October 2026.
49. Terms and privacy policy. Subtext. Privacy policy updated 10 October 2026. . Our own policy. Checked 13 October 2026.
50. Ovulation tracking app Premom will be barred from sharing health data for advertising under proposed FTC order. US Federal Trade Commission. 17 May 2023. . Regulator press release. Checked 13 October 2026.