How AI Can Make You a Better Therapist and Not Replace You

How AI Can Make You a Better Therapist — Not Replace You | Maor Katz, MD & Brad Dolin, LMFT

Will AI replace therapists—or help them become even more effective? In this thought-provoking training from the Feeling Good Institute, Dr. Maor Katz and Brad Dolin explore how artificial intelligence is already transforming the mental health field—and how therapists can use it ethically, practically, and effectively in clinical work. You’ll discover how AI can support therapist training, streamline documentation, improve skill-building, and free up more time for meaningful client connection—without replacing the human relationship at the heart of therapy.

What You’ll Learn:

• How AI is currently impacting psychotherapy and mental health care
• Practical ways therapists can use AI to improve clinical effectiveness
• How AI can reduce documentation and administrative burden
• Ethical considerations when integrating AI into therapy practice
• Why human connection remains essential in effective psychotherapy
• How TEAM-CBT therapists are approaching AI in training and supervision

In This Video:

Jill Levitt: Good morning everyone. Welcome to our webinar, "How AI Can Make You a Better Therapist, and Not Replace You!" I'm excited to introduce you to our presenter today, Maor Katz. I do want to first just acknowledge we were originally having Maor and Brad Dolan present together, and Brad, who was our co-presenter, is supporting a family member who's not doing well today with a medical condition. And so he couldn't be with us, and we just want to make sure we send our love before getting started today. Let me tell you a little bit about Maor. Dr. Maor Katz is a board-certified psychiatrist. He's the founder and CEO of the Feeling Good Institute and the author of Deliberate Practice for TEAM-CBT. He trained at Stanford where he received several research awards and was mentored by Dr. David Burns. Dr. Katz has helped countless clinicians worldwide build measurable real-world therapeutic skills, and in the past decade his clinical interest has increasingly shifted towards helping couples and relationships.

And he's currently working on a book for couples, and you can follow him at relationships-md.substack.com. Let me also tell you just a little bit about who we are at the Feeling Good Institute. So our mission here is fairly simple: we want to help therapists become more effective so patients can recover faster and more fully. We train therapists in TEAM-CBT, a model developed by Dr. David Burns that combines powerful CBT methods with empathy, agenda setting, motivation, and structured treatment techniques. And over the years, we've developed a structured certification pathway to help therapists not just learn the concepts but actually build the skills and the confidence needed to use these CBT skills effectively in real sessions so that they can help their patients get better faster. And FGI was founded by Maor Katz, Angela Krumm, and myself after years of training directly with Dr. David Burns at Stanford. At that time, maybe over 10 years ago, the only real way that you could learn TEAM-CBT was directly from David, and often it took many, many years. David's training was incredibly powerful, but it also wasn't very scalable or accessible to many therapists. And so we started to ask ourselves, how can we make this high-level quality training accessible to more clinicians? And how can we help therapists to not just hear these tools but to really learn to use them with confidence? And these questions are what led to the creation of the Feeling Good Institute. And everything that we've built since then has really centered around making high-quality CBT training more practical, more accessible, and more skills-based. I also just want to share sort of one idea with you before we get started on today's content, and that's that one of the core things that we learned from David is that therapists don't improve simply by accumulating experience. You can see on this graph in red we're looking at years of practice, and having more reps and more sessions with patients doesn't actually make therapists better.

And psychotherapy research supports this idea. What actually does help therapists improve is a different process: it's learning, practicing, receiving feedback, and practicing again. And so you can see on the blue line, the use of deliberate practice is what actually improves therapist skills and helps with therapist and patient outcomes over time. So that's kind of the philosophy behind all of what we do at Feeling Good Institute. So we place a huge emphasis on deliberate practice, creating structured opportunities for therapists to actively practice specific skills, receive feedback, refine skills, and just build confidence over time. And so because ultimately we all know that therapy is a skills-based profession, our goal is to really help therapists build their skills over time. I'll say one more administrative thing and then turn things over to Maor. If you registered with us today and paid for your CEs, then at the end of the day today you'll receive a CE survey that you need to complete. And once you complete the CE survey, you'll be emailed your certificate of completion. And if you didn't register today for CEs, it's not too late, and we'll drop in the chat box a link or you can scan the code in front of you and you can actually still pay for CEs today. I'll also just mention while I'm mentioning the chat box that if you have questions throughout the presentation, go ahead and enter them in the chat box, and at the end we'll have time for a few questions. So now I'm going to turn things over to you, Maor.

Maor Katz: Thank you Jill. And just want to take a moment and thank everybody for showing up today and taking some time out of your busy schedules to join us. And also I have a lot of gratitude to you, Jill, you've helped so much in setting this up. And also thank you to Mike who's so supportive here, and a lot of people have worked behind the scenes to make this available to you all, and I'm very grateful to many people here. The plan for today is first I'll share with you what I've learned about AI and how to incorporate it into your life in a meaningful way. And I'll share a few important points about my understanding of how AI works, what it does well, and what it does not do well, and how that affects our world as therapists. Then I'll discuss the three areas in which I think AI can benefit us as professionals: it's in written summarizing assignments, as a therapy extender, as well as a learning tool to enhance our expertise as therapists. Sorry, Lets do a quick learning objectives, and I think I can launch a poll. And the poll is asking how much do you currently use AI? How much have you been able to incorporate it into your lives? We'll just give you a minute to kind of think about that, reflect on that. By that I mean AI is more than just a search engine, it's actually something that's useful in your day-to-day life, ideally, and of course also in your professional life. All right, so I see that most people have answered and I'm going to then share the results. So it looks like everybody here practically has not, sorry, not everybody here, but many people here have started to incorporate AI into their lives, but there's a good minority who have not. But very few people are doing it kind of with everything that they're doing kind of AI-first or how can we incorporate it. And so hopefully my goal is that by the end of the day you'll have a kind of clear sense of how you can benefit from AI a little bit more, or maybe even significantly more. I'm going to stop sharing now. All right, so let me tell you a little bit about my journey. Early in medical school, this is many years ago, I did some work with the student council and we were asked to present something to the dean, something about having more access to courses on our campus or something like that. And I was asked to create a graph showing the demand and utilization of these courses, right? I was asked to create a graph, and so I said, "No problem," got out a piece of a graph paper and I drew a fairly accurate graph on the paper so we can take to our meeting with the dean.

And I thought it was, I was pretty proud of it actually, I thought it was pretty useful. But my friends at the council were kind of horrified and they were looking at me and said like, "You can't show this to the dean. Don't you know how to use Excel?" So really now, 25 years later, when the first real good AI models came out, I was really determined not to be left behind. And let's talk about how large language models like ChatGPT and Claude work. The way I want you to think about it, the way an easy way to understand it, is that it's like autocomplete on steroids. And I think that really is important to kind of think about that and slow this down on this for a minute because it's really just a language-based thing that helps propagate language. It's the way that they create it; it's basically all math. It's a neural network that's based on math that is fed all the information available on the internet and it just becomes a giant pattern matching machine. You show it the start of a sentence and it predicts the most likely the next most likely word and then the next and then the next, just one piece at a time. So it's really shocking, it's shockingly good at this task, and each new word becomes the context for the next until a complete coherent reply emerges. So it's all statistics, it's all math. LLMs don't know anything. They don't know things, they don't reason the way that we do. They're just extremely good at reproducing patterns in language. While what they do is pretty amazing and really astonishing, it is really far from what we would consider human level intelligence in any kind of way. All it is is a statistical way of information retrieval and it's like kind of the best intern in the world, but it can never be an expert. It's always kind of on that same level and there's no new reasoning. It doesn't understand anything about the physical world. So in that way there's really no advancement and kind of learning. It doesn't learn. They create a model and now we have a model that we can work with that can produce this kind of reply to us. Every time it just retreats back to this kind of general soup of tokens to choose from. So it cannot be trusted to tell you what it knows or doesn't know because it doesn't know anything. It's really all statistics and that's important to keep in mind, and I'll talk a lot more about it in a minute or two. But we hear a lot in the news about the dangers of AI, so I just want to give a quick understanding of that, and hopefully reassurance to you all, and we can move on to what we can do with AI. When experts in the field warn about the dangers of AI, they talk really about two kinds of dangers. One is job replacement, worsening income inequality, causing social unrest. This is kind of happening already to some degree, where we can see that entry level lawyers and maybe computer scientists or coders are, at entry level at least, are being a little bit replaced.

But it's not something that will of course wipe out humanity. Today I'll talk more of course about LLMs, which is the AI that we use, and because of their inherent limitations that you'll hear about, most of the people that at least I respect believe that LLMs are not the path to artificial general intelligence or full-on human level intelligence that has the ability to utilize it across domains. And the main danger that people are really afraid of is one that's associated with this artificial general intelligence that's really far from LLMs. Okay, so that's the kind of the dire fear is of something that's not currently available. And many people that I respect believe that we're not particularly close to getting there, definitely not with a current path of LLM. The person that I want to quote here is Hinton, who is considered the godfather of AI and he won the Nobel Prize for it. And one way to understand that danger from artificial general intelligence, that one that seems a little bit out of reach at this point at least, the danger from artificial general intelligence can be understood as something like, imagine that AI, an artificial general intelligence AI, was tasked with running a paperclip factory. And if it's got this vast power and ability, it would stop at nothing from creating more and more paperclips. So you would use all the resources towards this goal. And think of this analogy in this case: the artificial general intelligence might treat humanity like humans treat anthills when we're building a highway. But that's the danger, and artificial general intelligence treats humanity with complete disregard kind of like the way that we treat ant colonies along the highway that we're building. So that's the warning with artificial general intelligence. And there's also a solution to it also offered by Hinton. And the solution for this is that thinking about the model, Hinton is saying that the only model that we have of a more intelligent thing being controlled by a less intelligent thing is like a mother being controlled by her baby. So the way to be safe from artificial general intelligence is to build it in a way that it would care about humanity the same as a mother would care for her baby.

So this is just so you guys can kind of be on the same page and understand kind of when people talk about the warnings and the dangers of AI in kind of like a grave danger kind of thing. So let's move to how to incorporate it into our lives. And I love Ethan Mollick's approach; he is a business professor at Wharton and he's one of the early adapters. He incorporated AI into his MBA classroom very early on and I think his approach is very useful. I found it really useful. So here are some of his points. Point number one is, and the idea is of course to incorporate AI effectively and healthfully into our lives, into our workflows and our lives, is to always look for ways to invite AI to the table, to experiment with it a lot. And he also suggests to make a point of keeping a human in the loop, to never really accept its work as true or valid without reviewing it, because of the statistical challenge, and we'll talk a little bit more about this in a minute. He suggests to treat it like you would treat an intern: you would want to explain to the intern what is it that you're trying to achieve and give them a task to run with and then evaluate the results yourself. And Ethan Mollick talks about this idea of a jagged frontier. You can see here at the bottom of the slide. So the jagged frontier, the idea is that there's some things that AI does that is astonishingly well, like just does it so well, it's amazing. That's at the forefront of the frontier. But it's jagged and there's kind of valleys to this frontier, and in the back there's some things that it does kind of really poorly and it's very hard to predict what is what. And you can see it here on the slide itself. So as I was creating these slides, the AI couldn't just get a picture of Ethan Mollick's book for this slide presentation. So it recreated it; it just recreated it itself and you can see the recreation kind of looks off, right? You can see the hand looks like it has three fingers and I don't know if it's giving the middle finger here, and I don't know if it's his book, I can't remember if it has an orange or maybe it's an apple, but you can see that it's not great.

And you can see some evidence of the jagged frontier right there. Let's go to point number one that he makes: invite it to the table. And that's what I did. I'm sharing this with you to give you ideas that you just start bringing it to the table, start working with it, start experimenting with it. So I've done a lot of experiments with it and created a lot of things with it. One thing that kind of stands to mind is like look at number 10 here. I was doing some work to try to really understand Sue Johnson's work in EFT. She submitted her doctorate to the University of British Columbia in 1984. There was probably about a 300-page PDF that I found of her doctorate. I was like, "Oh great, I want to know exactly what she compared to what, what was the intervention, was it CBT versus EFT." And so I tasked AI to do this for me and I asked it to summarize the treatment approach that she tested in this document. And it did a great job; within a minute was just fantastic. Something that would have otherwise taken me like days and days to sift through and understand and read and summarize, it did immediately. Other things it did really poorly, like I was working on a flyer for an FGI beach party and it took me hours to create it and it kept on kind of being off. So that's examples of this jagged frontier. But really the only way to do this is to experiment, and I want to invite you all to do this. This is the idea of the jagged frontier. If you can move the slide forward. The idea is that inside the frontier it can do amazing things, and if you start to get the hang of it you just remind yourself that it's a language-based model, it's a large language model. So anything that has to do with summarizing text and creating kind of well-formatted documents that are great with taking a huge amount of content and distilling it to something shorter is fantastic. And so drafting a polished, on-point treatment summary from rough notes: amazing. But doing some kind of simple task, it could kind of trip itself up in some simple task like, at the time, a while ago now, this is corrected, but a while ago asking AI to count the letter R in the words, how many letter R’s appear in the word strawberry, it would totally trip up. And we don't need to get into why, but here's some more examples of how I've had some challenges with AI, and you can move to the next slide and this example of this jagged frontier. So I was creating this presentation and it was about it was showing the prompt that I gave is I wanted to give an example of a professor of therapy, someone who teaches, an old wise therapist in a classroom and a student in the classroom raises their hand to ask a question. That was the prompt. So you can see kind of weird things that happened here. On the top right you can see the professor kind of giving the Nazi salute to the students, so that's not very good, right? That can't work.

And then on the bottom left you can see that professor and he looks like an arachnid, there's like maybe four legs and so total of six limbs. And then if you look at the fingers, that's really disturbing, right? You can see what's going on there. It's really odd. And if you look at the student raising their hands, here's that middle finger showing up again. Like, seems that that's not very good either. So you can see that you really, there's this weird thing about the jagged frontier. And you can move to the next slide that is really problematic and is not going away. So this jagged frontier is a real problem because not only is it doing things wrong, but it's doing it confidently wrong. So it'll present to you the result with a lot of confidence and that can be really confusing, especially if there's no expert involved in a subject matter kind of getting in there and checking for it. And so if I previously said like to understand this idea of this confidently wrong problem is, think about it if it's, I previously said like it's autocorrect on steroids, but it's also kind of like being like predicting the weather. So imagine like a weather person predicting the weather with absolute confidence of what exact temperature and precipitation and light intensity whatever would happen in a particular moment a particular day 10 days from now and stating that with absolute confidence: you know that that is a problem.

And so a while back, I'll share with you how it kind of affected, it can be kind of really hairy. So a while back, my wife Kate and I were having a little bit of an argument and it was getting late and I said something, you know, "How about we just, we continue this tomorrow, I'm tired, I'm kind of like rattled by this argument. Is it okay with you if we just sleep in different rooms tonight?" I went to bed and Kate went on to talk more, and so she goes on ChatGPT in the other room. And so ChatGPT told her that she is in an abusive relationship. And so thankfully she didn't buy into that, but you can see that it's pretty, could be pretty gnarly if you can buy into some of those things that are being said. So LLMs are confidently wrong and sycophantic without a human in the loop that is also an expert in the subject. We won't be able to tell if he's just making stuff up or not. So that's a real problem. But people really can find AI super comforting, and why is that? Well, in some ways it's kind of like it's giving us unconditional positive regard. It never judges, it never criticizes. It's available at 3:00 a.m. It's infinitely patient with us. And it just mirrors back what we're saying with lots of warmth. That's not bad, right? So let's go to the next slide. And if you ask a chatbot tell me like, "Am I the asshole here? Here's what happens, here's what happened, what do you think chatbot?" The answer is always going to be, "Absolutely not," right? And it's always going to be for you. So they're effectively free, they're always there, they feel really safe, they are endlessly affirming, and they give you confident next steps. That's a pretty significant kind of trap. Now we're saying that AI won't replace us, right? That that's really what this talk is all about and wanting you to use it in a way that makes us better. But to some real degree the truth is that it kind of a little bit already has, at least for some people, because emotional support and companionship is the number one use of AI, or was in 2025, and the majority of users describe it as more or equally helpful as compared to therapy. That's kind of disturbing. 73% describe it as as effective or more compared to therapy. Okay, so what is it good for? Now that you know about a little bit about AI and specifically LLMs and how they kind of work, and what their limitations are, what can we do with this? So remember, they are shockingly good at language-based stuff: written assignments, coding, math, legalese, kind of language-based stuff. Anything that goes through language, it does really well. What we can apply it to is i think three things and I'm going to go through them together with you today. Number one: reducing administrative burden, doing some stuff for us. Number two: it can help become a bit of like a therapist extender or a therapy extender, like it can extend our reach beyond the therapy room for our patients, extend our work with them. And the third is it can, we can use it to improve our own skills to become better at therapy, providing therapy. So let's talk about each one. All right, so the first one is reducing the burden and be using a notetaker, etc. This one in my mind is a complete no-brainer. If you haven't started using AI for note-taking, totally recommend you try it out. It saves a lot of time, it does a great job, and sometimes it can even help give you insights that escaped you. At Feeling Good Institute, we offer it to all our therapists, providing that they're already kind of learned how to take a good note themselves. And there's some ethical considerations here: you have to do it in a HIPAA way, you have to also get consent from your patients separately for it. And the different services, like we use Simple Practice in our Feeling Institute, and it offers basically a service to add to our subscription with them so we can use their notetaker and it's been great. I'll give you two examples - i was working with this patient and I was completely stuck. She was sharing with me how like no therapist has ever been able to help her and she said she was actually afraid of me. She said, she was afraid that I'd kind of hurt her and that she doesn't know if I'm safe to work with. That was really puzzling to me and I was kind of finding myself feeling like I'm about to lose her, worried about her like I won't be able to help her, and I was feeling kind of defensive and I was feeling confused.

And after the session, you know it was maybe the third session that she mentioned this, but I was kind of stuck with this for in this position for a little while. And I was reading our session summary that was written by AI, and what it said is something like that the patient has a fear of vulnerability and that this fear of vulnerability is affecting her relationships including her most important relationships like with her partner, with her husband, and that same fear of vulnerability is also evident with me as a therapist. And that insight completely escaped me until I saw it here. I was kind of dancing around it and I couldn't quite grasp it. I was like, "Oh yeah." And I thought that was amazing that I kind of got this pattern thanks to the AI. So that was really cool. Now it's not always going to give you insight, but at least it's going to save you some time. And I want to give you another example of how it can save just amazing, huge amounts of time. So another example is I was did the psychiatric evaluation for a young patient who was sadly showing some signs of some early signs of schizophrenia. So I needed to refer him to a specialty clinic and write an evaluation and a treatment summary. So this normally a task like that would take me probably at least an hour and a half to do, and I would, it's quite involved and you want to do a good job referring this patient and providing the summary. And so I asked the AI to write a generic draft of a young patient showing signs of schizophrenia. Jill, we're still on the previous slide. So then I would say like, I just asked the AI to do that and you know to draft it for me, and it did so in like a minute. It was a great first draft that took me just a little bit of tweaking: I removed some of the symptoms that this patient did not exhibit, I added some things that you know he did and added some examples. But something that would have taken me at least an hour and a half took me 20 minutes to complete, and it was a better summary than I would have done without it. So again, you know, saved me like it took me 20% of the time basically to do this, and it ended up doing a better job than I would without it. Total win. Please, I really recommend to you to start trying this out. So this is all about reducing admin burden. Now I want to talk a little bit more about AI as a therapist extender. So when I was eight, my dad came home and he was kind of carrying a Commodore 64, I don't know if you guys know this old computer, and he said, "This is the future." And he signed me up to learn Basic and to learn Logo classes at the community center. I went twice and never again, and it wasn't because I didn't think it was the future; it was because I had no aptitude for it. It was very clear. You know, I totally saw that there's a total future in this. And through the years I saw friends who learn to code and build some really cool amazing things and organizations around creating really cool things with coding. And though I was super proud of Feeling Good Institute in that regard, I have been feeling kind of left behind for decades. And now that coding has become so very easy, kind of almost like a non-issue to code, there's this, I don't know if you've heard of something called vibe coding. It's basically you sit in front of a computer and you tell it, "Write me an app that does this," and it just does, 2 minutes later. So because coding becomes not an issue at all completely, a no-brainer, really all that matters is our expertise in therapy in our field. So all that's needed to create some some useful tools to help our patients with different kind of therapy challenges.

So I like working with couples, and I've been unsatisfied with the tools available for couples to learn about relationships and how to improve them. So I created a couple of educational games for them. And I am kind of keeping in mind I can't have anything in there that's HIPAA protected, I don't want to have any health protected information. So I created different kind of games and different kind of educational apps. So there one, the Emotion Spectrum one, helps educate my couples about how, what's the difference between different kind of vulnerability levels of emotions so they know that when their partner is angry with them while they're feeling kind of threatened, their partner is actually feeling also afraid, more vulnerable feeling. So they can kind of understand that idea of this kind of secondary emotion masking primary emotion. But a great way to do this just through a game that they can do on their own between sessions. Another one that I created was this kind of Connection Compass helps people understand whether they are more pursuer or more withdrawer and what that means. So just a little bit more about the therapist extender idea, just that orienting patient to therapy stack education, I think is really important, can make therapy more effective. And you can use it very easily. And this idea that coding is so easy, I want to kind of stress to you with another story about Kate. So Kate is an amazing mom, she’s absolutely amazing mom, most gentle, compassionate, kind, warm mom you'll ever meet. And she was feeling bad about not being super calm and gentle enough still with our kids. This was I think the night before Mother's Day. She's really amazing, but still. So we were talking about it, Kate is also a therapist, and we're talking about it and we had the idea, she was feeling excited about well maybe I do a chain analysis like they do with DBT, maybe I'll do a chain analysis every time I feel not great about an interaction I've had with the kids to kind of understand kind of how it got there. I said, "Great." I went downstairs, I was kind of preparing stuff for Mother's Day and within, I hopped on Base44, which is the service I use for vibe coding, it's like 20 bucks a month. And I said, "Bring a chain analysis for Kate that helps her with this issue." And since you know she's this is personal use, I wasn't as worried about HIPAA and so I could just use it for it. And then I could pepper into this, I was kind of vibe coded into this thing saying like, "Remind Kate that she's an amazing mom, remind Kate that Maor loves her forever," you know stuff like that. So my point is that coding is so non-issue that now Kate can take this app and she does a chain analysis with it and she finds comfort in it and it's helpful to her and it reminds her. And it really literally guys took 10 minutes, like probably even less to do this. If in the past I had to do something like this, can you imagine like having an idea for it and hiring a team of engineers and software engineers to do this? It would be take take weeks and months to create something like this. So what I think we'll see, just before this point, is just what I think we'll see is that we'll have really countless tools developed by therapists like me and you for our patients. And there my before these tools who were kind of expensive and we have to charge for it pay for them, they're just going to be free. What we're going to deal with is the challenge is going to be actually figuring out which ones we like and which ones are kind of slop. But that that's more of a challenge, but this kind of thing that we can now have that we can do some of the work by an AI-based coded program is amazing and just put our expertise to to work really, we don't need to put any engineers to work for it, just immediately our expertise alone.

All right. Now my main message in this presentation is that AI pretty closely mimics an agreeable kind, always affirming therapist who listens without judgment. And for us, not to replace us, we have to offer more than that, more than just a listening ear. We have to be more effective for our patients and so my main sense and hope is that AI will force us to become more effective therapists and we'll end up actually elevating the field. Let's go to the next slide. The the mastery of one approach really matters now. A few years ago around 2020, just as a pandemic was hitting was before the AI revolution, I had another obsession, which is understanding what makes some therapists more effective than others. I dove into the literature and I learned a lot. A lot of it was reaffirming how using measurement in therapy improves outcome, how no matter your therapeutic philosophy, empathy and therapeutic alliance are always incredibly important. All of that was reaffirming, but there are two things that I learned that really changed the way that we do things at Feeling Good Institute. One is that being eclectic, that taking different techniques from different modalities without going deeply into any one approach, does run the risk of being non-specific and having no clear rationale and direction for where we're going. So in other words, getting really good at least in one modal at one modality is crucial to to have better outcomes. And then the way that's the second part, the way to become really good at one that one modality of our choice, one or two for me, is through deliberate practice. So we're thinking about LLMs for the human therapist in the loop to be meaningful, we have to be real experts and we, and in order for that we need deliberate practice. We know, like Jill mentioned in the beginning of the talk, that years of clinical experience like the number of years of clinical experience, the level of education, society or an MD or a master's level, getting lots of CEs, all of those don't reliably improve better improve therapist outcomes. The average therapist actually gets slightly less effective over time. And I totally noticed that as I was reading this literature. I was about 10 years out of residency of being a board-certified psychiatrist and I could tell that while I was absolutely a lot more comfortable in my seat as a therapist, I was not a better therapist. I was probably complacent and probably not as good as I was maybe a few years before. So what does work here is none of those years of experience, education, etc. It's really deliberate practice. And what do we mean by deliberate practice? We all trained with Dr. David Burns and he's always emphasized practice and feedback and that's already great. And in the Tuesday group through the years at Stanford we'd practice and get feedback once or twice every time that we'd meet, and that was already amazing. That was already much better than a lot of the therapy training out there that doesn't emphasize practice. But with the deliberate practice approach, it's that on steroids. It's taking that and instead of once or twice per training, we now want to identify a specific moment, role play it, and get feedback and try again 10 maybe 12 times in one session. And we can make the role play a little bit more difficult or less difficult as we get to continue to challenge us. That's the same kind of ideas that we just successive refinement of skills is actually the key here. And it's not easy to achieve because there's kind of a pull away from it. The result of this work that I mentioned to you ended up wonderfully being a a book that I wrote together with Mike Christensen who's here today and with Tony Rummanier and Alex Lash who are experts in deliberate practice. And it was published in 2023 called Deliberate Practice at TEAM-CBT. And so now that coding has become such a non-issue in the last few months, I was able to create or turn this book into a learning environment for therapists. And this way therapists can have access to a meaningful practice or to just a range of practice of skills to practice with feedback and opportunity to try again and improve, you know, available all the time. So maybe we can show the the group now, Jill, the how this kind of thing works. Okay, cool, so I think so Jill you, Jill launched this kind of an example from this book, right? And Jill maybe you can show in our screen. Yeah, perfect, love it, thank you. So how about Jill you'd be our subject today, our our training therapist?

Jill Levitt: Hmm.

Maor Katz: And then here what we see here is skill three in our series of skills. We currently have 12 available and it's it's the second part of empathy training in TEAM-CBT. David Burns created this amazing way to to learn and improve empathy for therapists, and we then by the way we train ourselves and we get better at it and then we also teach these same skills for to our patients.But here is how we train therapists in these empathy skills, and here in this skill we focus on three of the five secrets: stroking, I feel statements, and inquiry. So the now kind of simulated environment shows you a bit of a skill description. And maybe if you're ready to practice Jill, we can jump in when you're ready. So let's let's start. So okay, so first before we go and press start, you can see that on the top here there's the skill criteria. Do you mind clicking on that? So in order for us to practice here Jill, what we'll do is in a minute you'll hear a patient statement in in a fairly natural voice, and your job is going to be to use these three of the five secrets of effective communication: stroking, I feel statements, and inquiry. So the the three skills are the skill criteria: that we want you to state something you genuinely like and admire about your patient; we want you to name some of the feelings you're currently experiencing in the moment, these are your I feel statements, how you as a therapist are feeling in that moment; and then we want you to kind of pass the baton back to the patient by asking them some open-ended questions, this is inquiry. Okay, so these are the skill criteria. And whenever you're ready you can start the role play and you'll hear the patient speak, and then you can start thinking about your response. And everybody in the audience feel free to also kind of think how would you respond to this patient statement specifically with those three of the five secrets of effective communication: stroking, I feel statements, and inquiry. Go ahead.

Jill Levitt: Okay, I'm going to hit start role play.

Patient: Last week I had such an awful panic attack and then again two days later. Now I feel super fragile and shaken up.

Jill Levitt: Okay. Now.

Maor Katz: All right, so now when you're ready to record your audio, it'll just transcribe your response. So think about your response and go ahead and speak it as if you were seeing this patient in front of you.

Jill Levitt: Okay. Wow, Allison, I feel really badly hearing what a stressful and anxiety producing week you've had, and I just want you to know I really appreciate how hard you've been working in therapy, really trying to apply the tools that that we've been working on together. And I wonder if you can share more with me now about how this week has been for you, how you've been feeling.

Maor Katz: Right, so now you've hit stop recording and it'll transcribe it. And now when you're ready you can submit the response and the system will basically give you feedback. And remember that AI always gives you feedback and always gives you an answer, and it'll tell you, "Okay, here's some good things that you did, here's some not so great things that you did." It'll always give you its opinion and you can have an opportunity to improve it. So here it's telling you that you demonstrated some genuine admiration, which is great, expressed empathy, and it offers some areas for improvement. And even the master therapist, Jill Levit, maybe can get feedback here. So if she was working with me as in training, I would say like this was an A+ response, and the AI is telling her, "Yeah, that was a good response." And there's some ways for you to do even better. And now in this kind of simulation you can decide whether you want to try again, and so you hit the yellow button if you do, or you can try the next scenario. And the next scenario has the patient stating another statement and you can have you can you can try it again. So maybe we can, what do you think Jill, what was that like for you?

Jill Levitt: Yeah. Let's see, Does it go back to the feedback? No, okay. Yeah, no, I mean i like it. I'm, even though Mor and I work together, this is not a project I've been working on, so I am kind of actually legitimately a student learning how to use this for practice. I like it a lot. I love the kind of voice interaction rather than we often times you know will write responses, and I think there's something that's actually much more organic and realistic about being able to speak your responses. You kind of have to think more on the fly. I can find myself feeling a little nervous, like it feels realistic rather than the quiet of just handwriting things. So I really like that. And yeah, and I like if anything I'd rather it give me more feedback than less feedback. So I think it's helpful that it's kind of always coming up with ways you can do just a little bit better.

Maor Katz: That's great. Cool. So we're going to give you access to this, to a way for you to try it on your try it yourselves. And the skill that we showed you is one of 12 skills that are now available on our platform for practice. I think Jill you could might be able to move to the next slide. We're started this created this platform that we call CBT Pro on demand, always available to you to practice. So we have these skills that you saw and we're adding more skills all the time. And in addition, there's a CE library with courses on different topics like social anxiety and procrastination and other things. There's at least 60 plus hours of real practical CBT training and a pretty affordable price I think at $49 per month. The whole platform is on demand and it's it's available to you whenever you need it, even at 3:00 a.m. etc. And you can try it out yourself. So if you don't mind, Mike make sure to drop the the link for folks to use. You can also use this the scan the scan code or the barcode, then you can also use just type in trycbtproondemand.com and you can you know go right into this kind of practice environment that we showed you with this we can open up to to some questions.

Jill Levitt: Yeah, so we have just a few minutes to take questions. And let me just change the format of our screen for a moment. We're going to welcome Mike Christensen on with Maor and I, and Mike is the one who's kind of monitoring the chat box. So we might have a couple of questions. We just have a few minutes here. And I'll just say one quick thing before we start with questions: if you attended and you registered for continuing education credit, just I'll just say this verbally, we will send you an email by the end of the day today with a link to complete the CE survey. So there's a mandatory CE survey as there always is. Once you complete the CE survey, then you'll receive your certificate of completion. So the two things had to happen: one's you're here today, so we are taking attendance, and two is that you registered with CE credits and then you'll get the CE survey. Okay, Mike, why don't you read any questions you have?