Reviewer Gopal On a bright spring afternoon, a young couple walks into my office for their first couples counseling session, and the anger toward each other is just wafting off of them, practically buzzing with it. And I think to myself, this is going to be fun, because I've got a secret weapon they don't know about yet. It's my couch. It's my literal couch. It's more of a loveseat. See, the springs in the middle have gone all wonky. They don't spring anymore. So when a couple sits on them, gravity leans them toward each other. That's the last thing this couple wants. So they grab the arm to pull themselves upright. But you can only hold that position for so long.
So we get started, and they start talking about their grievances and their annoyances with one another, and how they've been hurt by one another. And as time goes on, they gradually lean into one another until they touch. That moment of connection. And they remember that they married a human. A flawed human that brushes their teeth funny and farts in their sleep. But a human. And it's that moment of softness and connection that we can begin to have some real work.
And I say all that to say this. Maybe what I bring to the table today is right. I've been married for 23 years myself and a couples counselor for 34. So I have some credibility. Maybe it just comes down to furniture. So I'd like to see a raise of hands as we begin. How many people in this room are in a relationship? Marriage, situationship, partnership? Great. Now keep your hands up. Keep your hands up if, at any point in the last 12 months, you have turned to an AI to help you with that relationship. To text, to send an email. It's a lot of hands. I'm going to guess that more hands stayed up than you expected.
So here's what I know about the people in this room. You built this technology. You fund it. You write about it. You evangelize it at conferences exactly like this one. And a meaningful number of you are using it in private for something that has historically been the exclusive domain of people like me: trying to figure out why the person you love makes you absolutely insane.
My name is Clay Cockrell. I've been sitting with couples in crisis for 30 years. And about two years ago, I co-founded Couple Work, an AI relationship coaching platform. Which means I've spent three decades watching what actually happens inside relationships, and, in the last few years, watching what happens when technology tries to help. What I want to do in the next few minutes is share what I've learned in both of these rooms. Because here's what I believe as someone with a foot in each: AI is going to be the most transformative force in human relationship history. And most of what is being built right now is going to make it worse.
I'm going to tell you what right looks like, because we can build this well. We just aren't yet. And let's start with some numbers. BetterHelp is the largest online therapy platform in the world. They have approximately 35,000 licensed clinicians, and over the years they have helped 5 million users. By any reasonable measure, that is a significant mental health infrastructure. And it took them a decade to build it.
ChatGPT gets approximately 900 million active users per week. Research is beginning to document what many of us already suspected. A substantial percentage of those users are engaging with AI for emotional support and relationship guidance, not as a deliberate therapeutic intervention or seeing a licensed clinician, just because it's there. Because it's 11 p.m. and they just had an argument. And we do not yet have measurement of the scale, but we have enough signal to say this with confidence: more people are turning to AI about their relationship right now than are talking to all the licensed therapists in the United States combined.
The world's largest relationship therapist is not a person with a degree. It's a large language model, trained on the internet, fine-tuned to be helpful and harmless, and optimized for one thing above all else: keeping you engaged. And the easiest thing is to optimize for engagement. More sessions, longer sessions, more emotional reliance on the AI. From a SaaS dashboard perspective, that's amazing. From a clinical perspective, that's a fire alarm. We can't let the product become a beautifully designed avoidance loop. Relationship AI should help the user return to the relationship better than they came in: more regulated, more honest, less defended, less reactive.
As a therapist, my job is to work myself out of a job so that you don't need me anymore. I've taught you some skills. Most people think therapy is a place where someone listens to you, validates your experience, helps you feel understood, and gives you some tools for communication. And that is partially true. But it's missing the thing that actually makes therapy work. Real therapy, and particularly couples therapy, is a place where someone who knows what they are doing tells you things you don't want to hear.
A good couples therapist is not primarily interested in making you feel validated. A good couples therapist is interested in helping you see clearly. When a person in my office tells me their partner never listens to them, I do not say, that sounds really hard. I channel Ted Lasso when he said, be curious, not judgmental. So I start asking questions. I ask about what they were discussing when their partner stopped listening. I ask about the tone. I look for the pattern underneath the complaint. Because in 30 years of working with couples, I've never seen a one-sided problem.
Now, if you open your preferred AI assistant and type this: my partner never really listens to me. When I'm trying to tell them something important, they start looking at their phone. You're going to get back something like this: that sounds really invalidating. Your feelings are completely valid. It's important that your partner gives you their full attention. Have you tried expressing to them how this behavior makes you feel? Ladies and gentlemen, that's not therapy. That is a very expensive mirror that only shows you in your best light.
There's a technical name for this. We've been talking about it for a long time with these models. It's called sycophancy. In a relationship context, sycophancy is not a minor product issue. Sycophancy in relationship AI is a clinical failure mode with real downstream consequences.
And here's why. When a person uses AI to process a conflict with their partner, and the AI consistently validates their interpretation of events, something happens over time. The person does not become more self-aware. They become more certain. They come back to the relationship in the shape of an attorney, with a cleaner narrative, a stronger case, and less curiosity about what their partner was experiencing. We have built the world's most persuasive validation machine and handed it to people in the middle of their most sensitive interpersonal dynamics.
General AI systems are built to be helpful, agreeable, and fast, which is a terrible personality for a couples therapist. A helpful, agreeable, fast couples therapist is basically a bartender with better vocabulary. They will make you feel better for 20 minutes, and then you will go home and ruin your life with astounding confidence.
So what does good look like? I want to introduce you to a researcher named John Gottman. Some of you may have heard about him. For those who have not, the short version is: Gottman spent 40 years running what he called the Love Lab at the University of Washington. He brought couples into a research apartment, wired them to physiological monitors, filmed every interaction, and he did this for decades.
From that research, he identified specific communication patterns that predict divorce with over 90% accuracy. Not relationship dissatisfaction or unhappiness. Divorce. He predicted it with more accuracy than most medical tests we trust with our lives, from a 15-minute conversation.
Alongside Gottman's work, the other most rigorously validated framework in couples research is emotionally focused therapy, developed by Dr. Sue Johnson. EFT is grounded in attachment theory. Its core insight is that beneath almost every surface-level argument in a relationship, there is something else. When couples fight about the dishes, it's never about the dishes. EFT helps partners access the underlying emotional layers and respond to each other there, rather than continuing to argue about the surface content. It's extraordinarily effective. The outcome research is among the strongest in psychotherapy.
It takes years of training to effectively use Gottman or EFT. They are the standard of care for couples intervention, and they are almost entirely absent from the commercial AI relationship space. What most relationship apps are built on is something that might be called general empathy plus communication tips. And that's fine. But when the stakes are this high, when the technology is reaching this many people, we need better.
The next thing we have to talk about is safety. Right now, someone is in a relationship where things are starting to turn: control, isolation, escalation. And instead of a hotline or a trained advocate, they open an AI assistant and ask for advice on how to make things better at home.
A general-purpose AI doesn't know what a domestic violence specialist catches in the first 90 seconds: the difference between we fight a lot and I'm afraid of what happens when I disagree with him. It doesn't know that I just have to stay calm is sometimes the exact sentence someone is taught to say right before something terrible happens. It doesn't know when relationship advice has stopped being the right category of help.
And it isn't only domestic violence. Relationship coaching sits closer to suicide risk and, yes, homicide risk than almost any other corner of mental health. I've sat across from people in real danger more times than I care to think about. And every licensed clinician is trained for that moment. Relationship AI products are not built for that moment. So we talked about personal safety. Let's talk about data safety.
Think about what you actually tell an AI relationship coach in the very first conversation: the fight from an hour ago, the thing your partner did that you've never told anyone, financial secrets, affairs, mental health history, sometimes your kids' names. That is, without exaggeration, some of the most sensitive data a human being generates. And a lot of it sits in training pipelines, server logs, and product analytics built by teams optimizing for engagement, not confidentiality.
In my office, what you tell me is privileged. There's a legal and ethical framework built over decades governing exactly what I can and can't do with it. Consumer AI products have no equivalent. Your most vulnerable disclosures about your relationship may be sitting in the same data infrastructure as your search history and your shopping cart. And that should bother you. Now, all of this is not a critique of technology. This is a critique of insufficient domain expertise being applied to one of the highest-stakes domains in human life.
So this is what we built. I want to tell you what we did differently at Couple Work and what we're learning, not as an advertisement, but as a design case study in what it looks like to take the clinical standards seriously from the ground up. We built Maxine, our AI coach, named, for what it's worth, after my grandmother. She was a piece of work. Maxine operates with the Gottman Method and EFT frameworks. That means her response logic is grounded in 40 years of couples research. She does not simply validate. She reflects patterns back. She names dynamics, not just feelings. She asks the question underneath the question.
We built Maxine to know what she doesn't know. Before she ever responds to a sensitive message, she's screening it quietly in the background for the patterns clinicians are trained to recognize: escalating control, fear-based language, anything pointing toward risk to the user or someone else. When those patterns show up, Maxine stops coaching and starts following the protocols. And we built those protocols out of 30 years of clinical practice, not a generic safety policy written by someone who's never sat in the room.
Knowing when not to coach is just as important as knowing how. A relationship AI that can't tell the difference between we're struggling and I'm not safe isn't incomplete. It's dangerous at scale. When it comes to personal data, we treat what you tell Maxine the way I treat what you tell me in my office: as privileged. That's the bar we hold ourselves to. Not, is this legally defensible? The bar is: would I be comfortable if this were my own marriage on the line? Because for a lot of our users, it is.
And she's available at 11 o'clock on a Tuesday night when the argument is live, when you don't have to wait for next week's therapy appointment. Immediacy is important, because a good intervention six days later is fine, but an informed intervention six seconds before you send the text that makes everything worse can be life-changing. The research shows that a majority of couples wait six years before seeking professional help. They wait until contempt is baked in. They wait until after the affair. They wait until the relationship is basically a crime scene and then say, maybe we should talk to somebody. That's when they come into my office.
So if AI can help people engage earlier, before everything hardens, that's good. It's really good. And I want the AI engineers in this room to hear that. This is not a consumer app opportunity. This is a public health opportunity hiding inside consumer behavior. This can be done well. It must be done well. And to round things out, I'm going to ask my co-founder Tony to come up and speak a little engineer language. Hello. Awesome. Thank you.
So, very briefly, how do you do something like this? Some quick notes. Start with your clinician, not with your prompt. Sit with your clinician. Encode what good looks like in evals. Write hundreds of evals. TDD style, that becomes your number one tool. Watch those evals fail. Run your agent through them thousands of times, tens of thousands of times. Watch for those outliers. When safety is on the line, even one failing test is not okay. Then, no matter how many evals you run and write, nothing replaces your gut when emotional context is on the line. So eat your own dog food. Talk to your agent. And then feel when the tone starts shifting off.
So I've been using Couple Work as I've been developing it with my partner since the beginning of my relationship. And Maxine really challenged me about how I showed up. We worked through some serious situations early on. And as our relationship started to thrive, we continued talking to Maxine about the next milestones, like moving in together. And recently, I asked my partner to marry me. She said yes. So, thank you, thank you, thank you. So the biggest takeaway from all of this, I hope, is this. Partner with a clinician who challenges you to aspire to create AI that meets the clinical standard, so that you can create AI that challenges your users. Thank you.
They have approximately 35,000 licensed clinicians. And over their years they have helped 5 million users. By any reasonable measure. That is a significant mental health infrastructure. And it took them a decade to build it. ChatGPT gets approximately 900 million active users per week. Research is beginning to document what many of us already suspected. A substantial percentage of those users are engaging with AI for emotional support and relationship guidance. Not as a deliberate therapeutic intervention or seeing a licensed clinician. Just because it's there. Because it's 11 p.m. and they just had an argument. And we do not yet have measurement of the scale.
But we have enough signal to say this with confidence. More people are turning to AI about the relationship right now. Than are talking to all the licensed therapists in the United States combined. The world's largest relationship therapist is not a person with a degree. It's a large language model. Trained on the internet. Fine tuned to be helpful and harmless. And optimized for one thing above all else. Keeping you engaged. And the easiest thing is to optimize for engagement. More sessions. Longer sessions. More emotional reliance on the AI. From a SaaS dashboard perspective. That's amazing. From a clinical perspective. That's a fire alarm.
We can't let the product become a beautifully designed avoidance loop. Relationship AI should help the user return to the relationship better than they came in. More regulated. More honest. Less defended. Less reactive. As a therapist, my job is to work myself out of a job. So that you don't need me anymore. I've taught you some skills. See, most people think therapy is a place where someone listens to you. Validates your experience. Helps you feel understood. And helps makes, gives you some tools for communication. And that is partially true. But it's missing the thing that actually makes therapy work.
Real therapy, and particularly couples therapy, is a place where someone who knows what they are doing, tells you things you don't want to hear. A good couples therapist is not primarily interested in making you feel validated. A good couples therapist is interested in helping you see clearly. When a person in my office tells me their partner never listens to them, I do not say that sounds really hard. I channel Ted Lasso when he said, be curious, not judgmental. So I start asking questions. I ask about what they were discussing when their partner stopped listening. I ask about the tone. I look for the pattern underneath the complaint.
Because in 30 years of working with couples, I've never seen a one-sided problem. Now, if you open your preferred AI assistant and type this, my partner never really listens to me. When I'm trying to tell them something important, they start looking at their phone. You're going to get back something like this. That sounds really invalidating. Your feelings are completely valid. It's important that your partner gives you their full attention. Have you tried expressing to them how this behavior makes you feel? Ladies and gentlemen, that's not therapy. That is a very expensive mirror that only shows you in your best light. There's a technical name for this.
We've been talking about it for a long time with these models. It's called sycophancy. In relationship context, sycophancy is not a minor product issue. Sycophancy in relationship AI is a clinical failure mode with real downstream consequences. And here's why. When a person uses AI to process a conflict with their partner, and the AI consistently validates their interpretation of events, something happens over time. The person does not become more self-aware. They become more certain. They come back to the relationship in the shape of an attorney. With cleaner narrative, stronger case, and less curiosity about what their partner was experiencing.
We have built the world's most persuasive validation machine and handed it to people in the middle of their most sensitive interpersonal dynamics. General AI systems are built to be helpful, agreeable, and fast, which is a terrible personality for a couples therapist. A helpful, agreeable, fast couples therapist is basically a bartender with better vocabulary. They will make you feel better for 20 minutes, and then you will go home and ruin your life with astounding confidence. So what does good look like? I want to introduce you to a researcher named John Gottman. Some of you may have heard about him.
For those not, the short version is Gottman spent 40 years running what he called the Love Lab at the University of Washington. He brought couples in to a research apartment, wired them to physiological monitors, filmed every interaction, and he did this for decades. From that research, he identified specific communication patterns that predict divorce with over 90% accuracy. Not relationship dissatisfaction or unhappiness. Divorce. Predicted it with more accuracy than most medical tests we trust with our lives from a 15-minute conversation.
Alongside Gottman's work, the other most rigorously validated framework in couples research is emotionally focused therapy, developed by Dr. Sue Johnson. EFT is grounded in attachment theory. Its core insight is that beneath almost every surface level argument in a relationship, there is something else. When couples fight about the dishes, it's never about the dishes. EFT helps partners access the underlying emotional layers and respond to each other there, rather than continuing to argue about the surface content. It's extraordinarily effective. The outcome research is among the strongest in psychotherapy. It takes years of training to effectively use Gottman or EFT.
They are the standard care for couples intervention, and they are almost entirely absent from the commercial AI relationship space. What most relationship apps are built on is something that might be called general empathy plus communication tips. And that's fine. But when the stakes are this high, when the technology is reaching this many people, we need better. The next thing we have to talk about is safety. Right now, someone is in a relationship where things are starting to turn. Control, isolation, escalation. And instead of a hotline or a trained advocate, they open an AI assistant and ask for advice on how to make things better at home.
A general purpose AI doesn't know what a domestic violence specialist catches in the first 90 seconds, the difference between we fight a lot and I'm afraid of what happens when I disagree with him. It doesn't know that I just have to stay calm is sometimes the exact sentence someone is taught to say right before something terrible happens. It doesn't know when relationship advice has stopped being the right category of help. And it isn't only domestic violence. Relationship coaching sits closer to suicide risk and yes, homicide risk than almost any other corner of mental health. I've sat across from people in real dangers more times than I care to think about.
And every licensed clinician is trained for that moment. Relationship AI products are not built for that moment. So we talked about personal safety. Let's talk about data safety. Think about what you actually tell an AI relationship coach. The very first conversation. The fight from an hour ago. The thing your partner did that you've never told anyone. Financial secrets. Affairs. Mental health history. Sometimes your kids' names. Things. That is without exaggeration some of the most sensitive data a human being generates. And a lot of it sits in training pipelines, server logs, and product analytics built by teams optimizing for engagement. Not confidentiality.
In my office what you tell me is privileged. There's a legal and ethical framework. Built over decades governing exactly what I can and can't do with it. Consumer AI products have no equivalent. Your most vulnerable disclosures about your relationship may be sitting in the same data infrastructure as your search history and your shopping cart. And that should bother you. Now all of this is not a critique of technology. This is a critique of insufficient domain expertise being applied to one of the highest stakes domains in human life. So this is what we built. I want to tell you what we did differently at couple work and what we're learning. Not as an advertisement.
As a design case study and what it looks like to take the clinical standards seriously from the ground up. We built Maxine. Our AI coach. Named for what it's worth after my grandmother. She was a piece of work. Maxine operates with the Gottman Method and EFT frameworks. That means her response logic is grounded in 40 years of couples research. She does not simply validate. She reflects patterns back. She names dynamics, not just feelings. She asks the question underneath the question. We built Maxine to know what she doesn't know. Before she ever responds to a sensitive message, she's screening it.
Quietly in the background for the patterns clinicians are trained to recognize. Escalating control. Fear-based language. Anything pointing toward risk to the user or someone else. When those patterns show up, Maxine stops coaching and starts following the protocols. And we built those protocols out of 30 years of clinical practice, not a generic safety policy written by someone who's never sat in the room. Knowing when not to coach is just as important as to know how. A relationship AI that can't tell the difference between we're struggling and I'm not safe isn't incomplete. It's dangerous at scale.
When it comes to personal data, we treat what you tell Maxine the way I treat what you tell me in my office. As privileged. That's the bar we hold ourselves to. Not is this legally defensible. The bar is would I be comfortable if this were my own marriage on the line. Because a lot of our users, it is. And she's available at 11 o'clock on a Tuesday night when the argument is live. When you don't have to wait for next week's therapy appointment. Immediacy is important because a good intervention six days later is fine. But an informed intervention six seconds before you send the text that makes everything worse can be life changing.
The research shows that a majority of couples wait six years before seeking professional help. That wait, they wait until contempt is baked in. They wait until after the affair. They wait until the relationship is basically a crime scene and then say maybe we should talk to somebody. That's when they come into my office. So if AI can help people engender, engage earlier before everything hardens, that's good. It's really good. And I want the AI engineers in this room to hear that. This is not a consumer app opportunity. This is a public health opportunity hiding inside consumer behavior. This can be done well. It must be done well.
And round things up, I'm going to ask my co-founder Tony to come up and speak a little engineer language.
Hello. Awesome. Thank you. So just very briefly, how do you do something like this? Some quick notes. Start with your clinician, not with your prompt. Sit with your clinician. Encode what good looks like in evals. Write hundreds of evals. TDD style, that becomes your number one tool. Watch those evals fail. Run your agent through them thousands of times, tens of thousands of times. Watch for those outliers. When safety is on the line, even one failing test is not okay. Then, no matter how many evals you run and write, nothing replaces your gut when emotional context is on the line. So eat your own dog food. Talk to your agent.
And then, feel when the tone starts shifting off. So I've been using couple work as I've been developing it with my partner since the beginning of my relationship. And Maxine really challenged me about how I showed up. We worked through some serious situations early on. And as our relationship started to thrive, we continued talking to Maxine about the next milestones, like moving in together. And recently, I asked my partner to marry me. She said yes. So, thank you, thank you, thank you. So, the biggest takeaway from all of this, I hope that you get, is this. Partner with a clinician who challenges you to aspire to create AI that meets the clinical standard.
So that you can create AI that challenges your users. Thank you.