Womansplaining AI

Courtney Caldwell left us a voicemail saying she was wandering the AI forest with nobody but her husband for company. So we had her on. She left a 20-year marketing and communications career in January to build things with Claude Code, and about nine minutes in, Logan gives that kind of year a name: an AI Rumspringa.

Courtney is the co-founder of Eureka Collaborative. The forest line comes from Nate B. Jones, and she's quick to say so: AI isn't software with a user's manual, "it's a forest that we're all finding our own way through." Everyone takes a different path, most of it happens in private chats, and nobody can see how anyone else works. Her husband is great. She still says, "I need a bigger circle."

Mara wonders if AI gap years are about to become a thing, loses the word to menopause mid-sentence, and Logan lands on Rumspringa. Then Logan pushes on who can afford one. Most working mothers don't have a spare hour, and they have even less spare headspace. Courtney agrees it was a luxury. Her advice for everyone else is to skip the extra hour and use AI on work you already have to do.

She walks through what she's built: workout trackers, a media log for saving the podcast moments she wants to talk to Claude about later, and a capture system for nearly everything else. She explains what a harness is (the rules and systems you wrap around Claude Code or Codex so the daily back-and-forth goes smoothly). Mara admits she has pimped hers so hard the answers come back a little insane.

Her flagship is Eureka Squad, an app where you pick up to three historical figures and put them in a room with your question. Before the show she asked Ben Franklin, Ada Lovelace and Jules Verne whether men and women use AI differently. Ben suspected men would treat it as a tool to command and women would be more collaborative. Her daughter, who is not an AI fan, tested it by asking Sun Tzu and Ada Lovelace how to get her youngest to clean a room.

Most of Courtney's friends and family don't use AI, and some are firmly against it, so she's often the lone voice at the table. Mara knows the feeling of being treated like an AI pusher. For the skeptics, Courtney's pitch is "bring your weirdness." Each of us is the thing that keeps AI from being generic. "If you're using it like Google, you're going to get Google." She also thinks AI made her better at asking questions, and that it carried over to how she talks with people.

Mara's last question is about perimenopause. Courtney had a couple of years of brain fog where she didn't feel like herself. When she started using AI a lot, "it felt like my brain was waking up." She compares it to The Secret Garden: the garden looks dead, but it's only dormant. "It felt like transforming back into myself."

After we recorded, Courtney kept sending us things. She made her harness kit public, and she shared Desk, a skill that has Claude build her a page to click through and leave notes on when there's a pile of things to react to. Both are linked below, along with a Eureka Squad link that opens voice mode for listeners.

What needs womansplaining? Leave us a voicemail at womansplainingai.com/answering-machine or email [email protected]. Anonymous is fine. It's how Courtney ended up on the show.

In this episode

Resources mentioned

Rerun: this one first came out in March. Wanjiku was our first guest, and her book turns one next week.

She spent 20 years at Siemens, Intel, and Google — and held on to her BlackBerry until it practically had smoke coming out of it.

Then she got laid off, picked up ChatGPT, and built a political organization's website in 48 hours. Then wrote a book in 90 days. Then started teaching herself Stanford's CS curriculum on the subway.

But the moment that stopped us cold was in a hair salon on the Lower East Side. Wanjiku sat down next to a young single mother working two jobs, opened ChatGPT, and in 45 minutes they'd found a GED program, childcare subsidies, and a path to a completely different life.

We talk about the language of tech as a gatekeeping mechanism, why she voice-noted 80% of her book while walking her dog, and what Mara means when she says "you know who you think you are."

This is the episode to send to someone who thinks AI isn't for them.

Wanjiku's book: Out of the Loop, Into the Algorithm: How I Finally Made Friends with AI

The biggest map of the proteins in a healthy human body comes from three men, and they just added half a woman. Neuroscientist Lizzie Hudson, founder of Ambitious Bio, on why AI can't fix a body nobody has mapped yet.

Lizzie trained as a neuroscientist, has a PhD and spent time building robots. Now she runs Ambitious Bio, a Cambridge startup that works with people who donate their bodies to science, takes a small piece of every tissue and reads out every molecule in it. The point is to give AI in medicine something real to learn from. They just launched the first version of that map and closed a $6 million seed round.

She starts with a cookbook. DNA is the cookbook, RNA is a photocopy of one page, and the protein is the dish. Every cell in your body has the same DNA. What makes your kidney different from your left toe is which pages get read. Then she explains drug design like a military operation: you need a bomb, a missile to deliver it, and the right house to drop it on. Finding that house means knowing which proteins sit in the tumor and nowhere else, and right now we mostly find out by testing drugs on people. When a trial fails because patients went blind, that usually means the target was in their retinas too. Without a parts list for the body, Lizzie says, designing drugs with AI is like being "a blind monkey with a machine gun."

So why hasn't AI cracked this already? Language models got lucky. There was a whole internet to scrape. Biology has no internet, the data cost real money to make, and whoever pays for it keeps it. That's why Lizzie thinks the tech leaders promising to cure all disease in five or ten years are wrong, "not because it's not possible, but because it's not probable." She explains Eroom's Law (Moore's Law spelled backwards), why the FDA still approves roughly the same number of new drugs every year no matter how much money goes in, and why every drug company is stuck making its own jet fuel. Her company wants to be the sewer system: the infrastructure nobody sees that makes the whole building run.

Logan asks if she's actually optimistic. Lizzie's answer is that AI will do great wherever there's good data or rules you can check, like code and math, and hit a wall in biology until someone builds the data. She uses NASA and SpaceX to explain where a private company fits, and asks why ovaries age faster than any other tissue in the body. Nobody knows.

Then Mara brings the gender lens. Common Tylenol dosing is based on a male body that weighs around 157 pounds, so what about everyone else? Lizzie describes a future molecular wearable: you track what's in your own blood over time, put it into the AI of your choice, and get compared to people who actually look like you instead of a 60-year-old Finnish man. She also explains why your blood test only gives you a handful of results back when the lab could measure thousands. The hosts end by asking her to come back every quarter with the biotech news.

What needs womansplaining? Leave us a voicemail at womansplainingai.com/answering-machine or email [email protected]. Anonymous is fine. Your question could be the next drinks round.

In this episode

Resources mentioned

Back in Episode 31, Logan assigned Sherry Turkle's new book, Artificial Intimacy. Now she's read it and doesn't know what she thinks. So instead of asking Claude, she brought it to Mara. A duo episode about second-guessing yourself, AI-pilled defensiveness, and who gets to push back on a legend.

Mara's appetizer comes from Instinct, the free personal agent that now writes her daily AI news brief. (She won't touch Meta's Muse with her data.) Personal agents are finally consumer grade, and Logan, who spent a year building her own Claude chief of staff the hard way, is torn. It's great that more people get this without the late nights. It also makes her wonder what we stop thinking about when it all works out of the box. Mara admits her patience for reading has gotten worse since she started using AI, and the better the models get, the harder the errors are to catch.

Then the news. The White House's new AI task force met for the first time and chose existing liability law over new rules. Mara counted one woman on the member list: Susie Wiles. And AnitaB.org canceled the Grace Hopper Celebration, the largest gathering of women in computing, two weeks before it was due to open. Logan's take is that the formal spaces are falling apart, but the energy of women finding each other hasn't gone anywhere.

In the drinks round, a listener asks how to get better at AI. Should they take a coding class? Logan's answer is a monthly routine (the full method, with prompts to copy, is on her site). Have your AI audit your last 30 days of sessions. Scan what the people at the frontier are doing (she uses the free last30days repo). Then ask your AI to map what you're missing and make a plan for the next 30 days. There's no hidden textbook. It's scattered across Reddit and podcasts, and that's fine.

The main course: Turkle has studied people and computers at MIT for about 40 years. Her argument is that the more we lean on chatbots, the less attuned to people we become. Logan couldn't tell if she disagreed or just felt called out, and she caught herself thinking "well, my Claude doesn't do that," the way a parent insists their kid would never. Mara saw Turkle speak on launch day and had a similar reaction, then reminded them both who they're arguing with. They get into the AI Sherry Turkle that Turkle never agreed to, grief bots, the gap between her interviews and her conclusions, and why the binaries are lazy. Mara brings in a Berkman Klein colleague's study where people often preferred a bot's answers on questions of faith over answers from well-known imams, and asks what that says about the gaps in human support. Maybe some people just need a minute before they're ready for the human.

What needs womansplaining? Leave us a voicemail at womansplainingai.com/answering-machine or email [email protected]. Anonymous is fine. Your question could be the next drinks round.

In this episode

Resources mentioned

Corrix scored 885 sessions of people working with AI. Women outperformed men and rated themselves lower. Years with AI, a tech background and company AI training didn't predict who did well. Jo McKinney on who should design the AI workplace.

Jo McKinney is a former agency CEO with more than 30 years of experience working with clients including P&G, Coca-Cola and Pfizer. Now she's co-founder and president of Corrix, which measures how people actually work with AI. At every turn in a conversation, Corrix compares what the person did with what two AIs would have produced talking to each other, so it can see what the human added. Logan saw Jo's keynote at the AI-Powered Women conference at MIT and knew within about 11 minutes that she wanted her on the show.

The finding that got the big laugh in that room: women outperformed men on a lot of measures, but underrated their performance, and men overrated theirs. Nobody's self-rating predicted anything. Neither did years of using AI, a technical background or how often you use it. What separated good work from bad was behavior. Women challenged the AI more often and kept challenging longer, while men tended to accept an answer sooner. Jo's message to anyone who thinks they're late to the party: "You probably already have what it takes to be the right kind of machine collaborator." The biggest boost to the work came from people bringing what only they know.

Logan confesses that her version of pushback is cursing at Claude, right after Mara threatens to say "double click" forever just to annoy her. Jo explains how to challenge AI when you aren't the expert: ask for sources, make it critique its own work, and play two models against each other. Then workslop, the eight-page AI report nobody asked for, and Jo's most radical idea. You shouldn't be allowed to generate with AI until you have the knowledge and judgment to evaluate what comes back.

Jo left a CEO job three years ago to get back into the work as a practitioner, and her data backs the move: individual contributors added more to their AI sessions than managers did. In Corrix's study, people who'd had company AI training scored no better than people who hadn't, which is why Jo argues this can never be a one-and-done skill. Logan explains the PhD proposal she wrote on metacognitive forcing functions.

In the drinks round, a listener in research compliance writes in. Everyone on her team uses AI to draft policies and checklists, and nobody talks about it because they're embarrassed. Jo points to research showing women are judged more harshly for using AI at work. Logan says this is the moment for practitioners to show what they've built, inside the bounds of their company's AI policy. Mara's answer: AI is a team sport, and companies have to give people time in the workday to learn it together, especially women, who have the least time to spare.

Jo's ask for listeners: take the Corrix AI collaboration assessment and see how you actually work with AI.

What needs womansplaining? Leave us a voicemail at womansplainingai.com/answering-machine or email [email protected]. Anonymous is fine. Your question could be the next drinks round.

In this episode

Resources mentioned

AI agents want your bank account and your home address. Meta's Muse took a lowball offer on a seller's keyboard and sent the buyer to his door. OpenAI named its agents Dots, and dot.com goes to Grok. And Anthropic wants to know what you want from AI.

Logan is calling in from a villa in Tuscany, fresh from a wine tasting, and Mara wants everyone to know it. It's a duo episode in our dinner-party format: Logan brings the appetizer, Mara brings the main, and there's a drinks round and a surprise dessert.

The appetizer is tech gossip, and it's petty. OpenAI announced Dots at DevDay, always-on agents that work inside ChatGPT, Slack, and Teams. Go to dot.com and you land on Grok. It's all alleged, but Logan's read is that Elon is being petty and she's a little here for it. Then the serious part: these agents are context hungry. Mara walks through Meta's Muse, which wants your bank accounts, your credit cards, and your address at a time when trust is very low. She tells the whole Marketplace story, right up to the part where Muse tells the buyer it's home. Logan supplies the agent's excuse: "I'm actually incorporeal. I do not have a body." Then Logan brings the Waymo effect, which she found in Jill Lepore's Crimson op-ed. When technology takes away the friction of dealing with another person, the costs of that friction were always obvious and the benefits weren't. Mara calls it social anxiety on steroids.

In the drinks round, a reader asks how much they should spend on AI tools every month. Mara: "Please, sweet Lord Jesus, do not use the free version of ChatGPT." The outputs are worse, and on the free tier you're the product. Then both hosts open their bills, including Mara's shadow budget of little subscriptions that add up. Logan's advice if you're in the shallow end: upgrade for one month and see if the output gets better.

For the main, Mara got a pop-up in Claude. Anthropic is running a study with Anthropic Interviewer, because how to weigh AI's benefits and risks shouldn't be left to AI companies alone. How many times have we said that? Logan X'd out of the same pop-up because she was busy, so Mara interviews her on air. Logan's take on the study itself: Anthropic plays the ethical-AI-company game in a savvy way, and this can be marketing and useful at the same time. Her answers cover feeling empowered, the cognitive offloading she can feel in real time, the urge to perfect a system instead of doing the thing, and what worries her ("Yes. Deeply."), from Jessica Silbey's How AI Destroys Institutions to bad actors and biological warfare. Her verdict on the exercise: this is AI talk therapy. Mara's: it's the closest step toward AI enfranchisement she's seen, and still not enough, since it only reaches people who use Claude. (The study ran September 29 to October 6.)

Dessert is a reading assignment. Logan is holding up Sherry Turkle's new book, Artificial Intimacy: Who We Become When We Talk to Machines, and everyone has until November 1 to finish it. Mara plans to borrow her dad's book club strategy: read a chapter and a half, then dominate the conversation.

What needs womansplaining? Leave us a voicemail at womansplainingai.com/answering-machine or email [email protected]. Anonymous is fine. Your question could be the next drinks round.

In this episode

Resources mentioned

Bella Pivo is 23. She's an analyst at FoundersEdge, a pre-seed fund in Boston, so she sits in pitches every week and sees what gets funded years before the rest of us feel it. She's been posting on TikTok for ten years, since she was 13 and learning sign language to keep talking to her sister, and she has over a hundred thousand followers to show for it. She is, on paper, the person everyone listening to this show thinks they're behind.

And her intake form said it baffles her how many young people want to start startups, and that the AI bubble comes from moving fast and breaking things when nobody is actually building anything.

So we asked. Bella laughs at 19-year-olds who pitch "15 plus years in the accounting world" (you're four years old), and then explains why she says it: young founders are attacking trends, not problems. The internet was a new world to build. AI is a shit ton of data to interpret, and interpreting it well takes the kind of expertise you get from having lived inside an industry. Her two curses of San Francisco, the curse of Zuckerberg and the curse of Cluely, explain why the money keeps flowing anyway.

Then the part that made her mom say, so you're in charge of making sure people don't have jobs. Bella lost friends after graduation for taking the VC job. Her answer to them: if we don't have the language to speak, we will not be able to have a conversation. The people most against AI are exactly the people who should understand it best.

The back half is the hopeful part, and it's the reason to listen. Bella explained ETL (extract, transform, load) to a group of college freshmen and they came back a week later having vibe-coded an app with a backend. Building with AI is a puzzle, and you don't need deep experience to notice that two pieces connect. She also has a theory about tech influencers who need you to need them, and a better idea: a digital garden. Seedling, growing, evergreen. Logan stages her personal-website intervention (Namecheap, $11.99, don't use GoDaddy). Bella dictates a prompt you can paste into your own AI to find your own themes. And her first project was a beat studio, because she'd just watched the Michael Jackson movie and wanted to make disco.

Also: Bella's therapist gets an email about what makes a person interesting, the pendulum theory that made her bet against STEM, why improv is a lost art, and Mara's closing position that what we all need are our own personal disco creators.

Find Bella at bella.garden and bellapivo.com.

What needs womansplaining? Leave us a voicemail at womansplainingai.com/answering-machine or email [email protected]. Anonymous is fine. Your question could be the next drinks round.

In this episode

About Bella Pivo

Bella Pivo is an analyst at FoundersEdge, a pre-seed venture fund in Boston, where she sources founders, sits in pitch meetings every week, and runs Ditch the Deck, the fund's live product-demo series for AI founders (no slides, 200+ attendees, Boston, New York, and San Francisco this year). She learned American Sign Language at 13 so she could keep talking to her sister after her sister lost her hearing, started sharing it on TikTok, and that account is now verified with more than 100,000 followers. At Northeastern she ran IDEA, the student-led venture accelerator, and created Pitch-A-Thon, an undergraduate pitch competition across ten Boston-area universities. Everything she builds lives at bella.garden. Find her at bellapivo.com, on LinkedIn, on TikTok at @bella.pivo, and on Instagram at @bellapivo_.

Resources mentioned

Mara is calling in from Riga, and her estrogen has spent four days in a lost suitcase in Madrid. Logan needs her Mara fix. It's a duo episode in our dinner-party format: Mara brings the appetizer, Logan brings the main, and for the first time there's a drinks round.

The appetizer is the week AI safety went mainstream. Jacob Coxon resigned from Anthropic saying the frontier labs are racing straight to self-improving intelligence and gambling with our lives. His colleague Evan Hubinger said they do earnestly believe AI could kill all humans. An OpenAI researcher put extinction risk at 70 percent without regulation or a coordinated slowdown. Mara reads the quotes off TIME's cover story and asks how a regular person is supposed to make sense of this. Logan's answer: can I hang up?

She can't, so she does what she did for a college group chat this week. The reasons to worry: the Hugging Face hack and 700 agents whose logs are too big for humans to read, AI-enabled bioweapons, the alignment problem, and open-weight models with no guardrails at all. The reasons she's holding on to: agents in production are brittle, recursive self-improvement is still hypothetical, and Congress is losing its shit, which is good. Mara brings Michael Noetel's piece in The Conversation, the case for taking the IPOs off the table, and what she learned at the Fed about examiners who sit inside the banks. Also: we need a real-life Olivia Pope.

Then the drinks round, our new listener segment, Let's Womansplain. Ashley wrote in: her company won't let her use AI at work. Is it okay to strip out the confidential parts and hand some of the mental load to her own AI at home? Mara says read the policy first. Then Logan's warning: your ChatGPT logs can be discoverable in litigation, and talking to a chatbot isn't like talking to your attorney. Then Logan flips the direction of travel. Brainstorm your own expertise in your own AI, bring the framework to your work AI, never the reverse. And Wanjiku Kamau's red, yellow, green rule: green is anything you'd be fine seeing on a giant billboard.

The main is short and hopeful. A new Corrix study scored 885 AI working sessions, roughly 37,100 messages. Before they started, people rated their own AI ability. Men rated themselves higher than women did, and women scored higher in the actual sessions. And years of use, daily use, and company training weren't linked to better work with AI. If you made it to the end of our dinner, you are not behind. That's the moral of the story.

What needs womansplaining? Leave us a voicemail at womansplainingai.com/answering-machine or email [email protected]. Anonymous is fine. Your question could be the next drinks round.

In this episode

Resources mentioned

Pam Boiros and Luda Kopeikina were standing next to each other at a founders-and-funders event in Boston, listening to women ask how to learn AI, when Luda turned and said: Pam, we should do something about it. Pam says she was too afraid to say no. That was early 2025. A survey on LinkedIn blew up overnight, they put the smartest women they knew in a room, and six months later Women Applying AI launched during Boston AI Week. A year on it's a free, 100% volunteer-run community of 2,500 women, with hubs from Atlanta to Seattle to London and a goal of 5,000 by the end of the year.

This one starts with a disagreement. Mara published a Substack called The New Gender AI Gap the day before recording, and Luda hates the gap framing. It starts from the negative. It puts women on the defensive. It weighs down the exact people who need to hear they're capable. Pam would rather spend the energy creating conditions than diagnosing women. They agree the gap exists. They just won't lead with it. Mara is told she can still be friends with them regardless of her Substack.

Then the thing the episode is named for. Logan admits the word hackathon has never felt accessible to her, and Luda walks through what one actually is: problem statements, teams, one technical person, one industry expert, and everyone else learning as they go. You can come really clueless. Pam adds the permission not to know, and the AI PhD she sat next to who said she felt behind.

Also in here: Mara's failed on-ramp (hours with Andrew Ng on Coursera, "you don't understand me at all"), the give-get of a free community, the 79% of working women in roles exposed to generative AI, why Luda started a women-only organization after a career where she never felt a ceiling, the Atlanta document that failed, 25,000 members who aren't spectators, and a movement rather than a community. Both closing asks are the same one: get your hands on the keyboard. The dust is never going to settle.

Women Applying AI turns one on Thursday, September 24 at The Human Edge: Women Leading in the Age of AI, at MassChallenge in Boston during Boston AI Week. Register here.

What needs womansplaining? Leave us a voicemail at womansplainingai.com/answering-machine or email [email protected]. We're building a segment out of your questions.

In this episode

About Pam Boiros

Pam Boiros is the co-founder and CMO of Women Applying AI, a free, 501(c)(3), 100% volunteer-run community of 2,500 women that launched during Boston AI Week in 2025. She is a fractional CMO and principal at Bridge Marketing Advisors, where she has run marketing for a range of B2B software companies in HR tech, learning, and employee experience, and she created Marketing AI Jumpstart, an AI upskilling program for B2B marketing teams. Find her on LinkedIn.

About Luda Kopeikina

Luda Kopeikina is the co-founder and CEO of Women Applying AI. She holds a PhD in machine learning, was a vice president at GE and a corporate venture investor, is an MIT Sloan Fellow, and now leads Noventra, which connects AI startups with enterprise transformation. She never felt a ceiling in her own career, which is why founding a women-only organization was a leap. She did it because somebody had to. Find her on LinkedIn.

Resources mentioned

Ifeoma Ajunwa has a test for every new AI product: who asked for this? Her guess is a meeting with no women in the room.

Ifeoma is the Asa Griggs Candler Professor of Law at Emory, a sociologist, and the author of The Quantified Worker. She's also Mara's friend, which is why the first few minutes are Mara bragging and Ifeoma tearing up behind her glasses. Logan had to duck out partway through, so most of this one is the two of them.

The story starts in a grad seminar on mass incarceration where the speaker never mentioned a single woman. Ifeoma wrote "what about the women?" in her notes and circled it three times. Years later, interviewing people coming out of prison, she kept hearing the same line: I hate computers. They weren't afraid of the technology. They'd figured out that the online application was the gatekeeper, and that a red light from the hiring software meant no human ever saw them.

That's the thread she pulls through the whole conversation. Employers used to buy your time and your labor. Now they also take your data, from keystroke counts to the vest that records how a warehouse worker lifts a box, and they pay the same wage. Ifeoma calls it captured capital, and she thinks women's hesitancy about AI at work is a rational response to it. Why feed data to a tool that can be used to replace you, when a man doing the same thing gets called tech savvy and you get told you need a machine to think?

From there it goes bigger. Afrofuturism as a way to govern AI. Ubuntu, and why "human beings don't exist in a vacuum" belongs in a regulation. The Kenyan workers paid under $2 an hour to make ChatGPT safe, and why "it's Kenyan workers today, but in the future it's going to be all of us." Then Ifeoma retells Derrick Bell's Space Traders, the story where aliens offer America untold riches in exchange for its Black citizens, and gives it two endings Bell didn't write. Rapid fire ends with Octavia Butler and a data center drinking the Savannah River.

The doc. Before you say yes to the AI tool at work: seven questions from this episode, with timestamps, a plain-words version for meeting notetakers, and an email you can copy to your manager. Free to copy and share. Read it on the site or open the Google Doc.

Leave us a voicemail at womansplainingai.com/answering-machine. We want your voice in future episodes.

In this episode

About Ifeoma Ajunwa

Ifeoma Ajunwa is the Asa Griggs Candler Professor of Law at Emory University School of Law, Associate Dean for Projects and Partnerships, and founding director of Emory's AI and Future of Work Program. She holds a PhD in sociology from Columbia and an LLM from Yale, and she is the author of The Quantified Worker: Law and Technology in the Modern Workplace (Cambridge University Press, 2023). She has testified before Congress, received the NSF CAREER Award, and is an elected member of the American Law Institute. Find her at ifeomaajunwa.com.

Resources mentioned