Claude Fable 5 Asked Me The Questions My $300/Hour Coach Used To

Claude Fable 5 Asked Me the Questions My $300/Hour Coach Used To

Claude Fable 5 has been everywhere on your feed these past weeks, with AI gurus swearing it was your last chance to try Anthropic’s smartest model before July 7th. I saw the same countdown timers and red arrows you probably did, and before I said a word about it to my own community, I had my AI fact-check the claim first. What I found changed the whole story I was about to tell, and it also handed me a simple way to decide when the expensive model is actually worth using.

In this episode of Run Your Private Practice with AI, I run five real tasks from my business through Claude Fable 5, from a Facebook post to an offer audit, to find out where the top-tier model actually earns its keep. I walk through what surprised me, including the one question that will save you from overpaying for the wrong model ever again.

AI Tips You Don’t Want to Miss:

Why a weekly Facebook post does not need your most expensive AI model, and what tasks actually do.
How a two-minute fact check exposed the real story behind a viral guru claim.
The way Fable spotted a pattern across 100+ podcast episodes that I had never connected myself.
How a “poke holes in my offer” prompt surfaced the same questions a $300/hour business coach used to ask.
The simple test I now use to decide if a task is worth the expensive model or not.

Mentioned in this episode:
Run Your Private Practice with AI (free Facebook group)
https://www.facebook.com/groups/aiprivatepractice/
Clinical AI Club: https://www.clinicalaiclub.com
Thera AI Hub: https://kymtolson.kartra.com/page/AI-Tools-for-Therapists

Connect with me:

Instagram: @thetravelingtherapist_kym
Join the Clinical AI Club: www.clinicalaiclub.com

Sponsored by Alma: www.helloalma.com/kym
Sponsored by Berries:www.heyberries.com/therapists
Sponsored by Auralink Pro: Start your free trial and use code KYMTOLSON for 20% off www.auralink.com/pro

Transcript:

[00:00:00] Hey, y’all. Welcome back to Run Your Private Practice with AI. I’m Kim Toulson, and I’m recording this episode basically from my bed. I actually just had surgery, and it was way rougher than I expected. I thought I was gonna have surgery, get up and go and be fine. I’m used to really just pushing through because honestly, I’ve got a lot of chronic health issues, and that’s just what I do.

[00:00:21] I just continue. I just keep pushing through. So I’m starting to accept the fact that recovery is gonna take longer than I planned for. So if I sound a little different today, that’s why. I’m really propped up on pillows here with a mic, and I’m just gonna record this episode ’cause I like to have an episode every single week.

[00:00:39] And before I get into the actual topic, I really wanna say just a little something about that because it’s really the whole reason this episode exists at all. A few years ago, an unexpected surgery would’ve been a business emergency for me, especially just in full private practice. It would be canceled sessions, loss of money, emails piling up, launches or ideas just dying.

[00:01:03] But right now, my newsletter went out on Sunday. My social media posts are already scheduled. My team knows what they’re doing. My AI systems are running the stuff that used to eat up all of my afternoons and my evenings, and the only thing I actually have to do this week is lie here and heal, which is hard of anything for me, and record this episode apparently, because I can’t help myself.

[00:01:26] And my podcast editor just messaged me and said, “Where’s your episode from this week?” And I was like, “Ah, crap. I didn’t do it yet.” So there’s so much going on always in AI, it’s hard to even pick a topic. If you heard the last episode 106 with Sally Riggs about using AI to run your business without a team, this is that idea in real life.

[00:01:46] So I’m living the case study right now from my bed. If you’ve been putting off building your systems because things are fine at the moment, let this be your gentle nudge Things are fine until they aren’t fine, and you want systems built before the surgery, not after the surgery or on the fly when you find out you have to have surgery, so to speak.

[00:02:08] So on to what I actually wanna talk about now. So while I’ve been lying here trying to recover, I’ve been scrolling a lot. I’ve been taking painkillers. Painkillers and Instagram are really a dangerous combination. I even went into, like, a emergency coaching session with painkillers, and I was like, “This was a terrible idea.”

[00:02:27] She was like, “Yeah, we can deal with this next week.” I was like, “Okay, probably a good idea.” But anyway, so my whole feed this week, and if you follow AI at all, you’ve probably seen it, like, every AI guru I follow and several that I don’t, they were screaming the same thing, “Use Claude Fable before July 7th while it’s still free.

[00:02:45] This is your last chance,” with red arrows and shocked faces and countdown timers. You’ve probably seen the post if you are following AI at all. Fable 5, if you missed it, is Anthropic’s newest model. It’s the one that sits above Opus now, so it’s their smartest model. In fact, the government actually made them take it down for a little bit because it was so powerful and they were, I guess- Freaked out about everything it could do, supposedly autonomously.

[00:03:10] Now, I’m not a coder. If you’re a coder, I think this is way different, but for me and somebody probably like you, I don’t know, I wasn’t that impressed with it. And like right now there’s this promotional window, it’s ending on July 7th. It’s July 6th right now when I’m recording this. By the time you’ll hear this, it’ll already be past.

[00:03:27] There was like this promotional window where paid Claude subscribers could use it as part of their plan before it switched over to like a pay per use credit system. Now, by the time you’re hearing this, like I said, that window is probably already closed, and I want you to keep listening anyway because I promise the point of this episode is not like that deadline or trying to give you urgency to go rush and try it.

[00:03:51] The gurus out there make it sound like it’s the end of the world if you don’t go do this and test it before this deadline ends. And here’s what my brain did, and yours might too when you see posts like this. I’m like, “Okay, it’s free. Is it actually free? Like, I’m already paying for a paid subscription, so how’s it free?”

[00:04:06] So I did what I always do now, and this is a good workflow for anybody. I asked Claude to fact check the claim before I posted anything about it to my own community out there. And the gurus, of course, were half right and half like just, you know, clickbait titles, which for the gurus is honestly a pretty good batting average.

[00:04:25] That’s pretty much what you can expect with most of those headlines. You click in and it’s not exactly what the headline says, of course, but I understand needing the clickbait type thing because really, I mean, YouTube, Facebook, LinkedIn, Instagram, all that, nothing even gets acknowledged and if it doesn’t have some kind of like hook baity type title with it.

[00:04:45] You know what I mean? So even I feel gross doing it sometimes, but then it’s like you have to if you wanna even like show up in the algorithm. It’s kind of a whole… I should do a whole episode on that. And the comments that people make when I do the clickbaity things. Like just the other day, I, you know, I was not feeling great.

[00:05:01] I posted something, this kind of clickbaity thing in my Facebook group, and somebody said something like, “The scammy dishonesty in this post is sickening.” It was a therapist who posted this, of course. Therapists are the worst in my opinion. No offense, but really if you’re in any Facebook group, you’ve seen it yourself probably.

[00:05:19] posted that and somebody comes in right under it and is like, “Yeah, this page needs a vibe check,” or something like that. I’m like, oh my God. And remember, surgery, painkillers, you know, and I was like, excuse my language, I was like, “F these people.” So I just booted them out of the group ’cause I’m, I’m just so sick of like the comments and the, you know, need to like even comment something like that.

[00:05:41] It’s like if you don’t like it, just move on. It’s the way you play the game, so to speak, if you want to have a successful social media channel. ‘Cause under that, you know, after their dumb comments, there were like six to 10 messages of people genuinely saying, “Oh, I want more information about this.” So yes, you have to use the clickbaity thing, but then there’s actual real value in a lot of this stuff.

[00:06:04] So the Fable 5 thing, it was included for paid plans, Pro, Max, through July 7th. Not free accounts, paid plans, and you could only use it for up to half of your weekly usage limits. So the whole like, “Free for everyone, this is your last chance,” was really, it’s included with your existing subscription capped to 50% for one week.

[00:06:23] So this happens a lot with everything, right? You have to read the fine print. Anyway, so here’s the part that killed the urgency completely for me. Anthropic came out and said that Fable’s expected to come back to subscriptions once they have enough server capacity for it. So this was never really like a last chance anything.

[00:06:39] It was more like a preview of Fable 5 so you could see how good it is. So I’m not telling you this to like, lay crap on anybody’s posts and that sort of thing, or even Anthropic and Claude. I’m, I’m telling you because the fact-check took me about two minutes with AI, and it changed everything about what I was gonna say publicly about it.

[00:06:59] So first I was like caught up in the hype. I was like, “Oh my God, I’ve gotta figure out everything Fable 5 can do for therapists and give them a, a rundown so they can go do it before July 7th.” So anyway, of course I wanted to go down the rabbit hole and see if I could like find some really cool things to tell you all about.

[00:07:14] Since I had the window where I was in bed doing nothing except for staring out the window and had time on my hands, I decided to actually test Fable and see if I could get some really cool stuff. So not the, “Ooh, look at the shiny new model” testing, but like real testing. So I took five actual tasks from my actual business and ran them through Fable to see what all the fuss was about.

[00:07:34] The five tasks were writing a Facebook post for my community, fact-checking that post, that guru claim that I was just telling you guys about, and then coming up with this week’s podcast episode, which I already do. Pressure testing a new offer, an idea that I have, and then auditing my intake paperwork for gaps to see if I had any gaps, ’cause I saw that a lot of people were saying, “Use it to audit everything while you can,” that kind of thing.

[00:07:57] So I did that. And I wanna tell you what happened, because the results actually surprised me, and they’re probably gonna save you a lot of money. So half of those tasks did not need that fancy model at all, the Fable model. So the Facebook post, because this is the one that made me go, “Wait a second,” I did have Fable write a post for me for my community about Fable and this whole research I was doing.

[00:08:18] And funnily enough, the Fable promotion itself So five ways therapists could test the model before the window closed was like my hook, and the post was good. It sounded like me. The ideas were solid. It was accurate because I’d already done that fact-check first. But here’s the thing. I, I’ve been writing posts with Opus and Sonnet and Haiku for months, and when I looked at that Fable post, I could not point to a single thing that Opus wouldn’t have done the exact same.

[00:08:44] The same voice, the same quality, the same everything, and I literally sat there and I was like, “Did I just use the expensive model for this post?” And the answer is I really shouldn’t have. A social media post is a couple hundred words and a voice the AI already knows about me because I’ve trained it on my brand guide, and there’s just not enough, like, need for the extra intelligence to show up in a post like that.

[00:09:06] It’s like hiring a Michelin chef to make you a piece of toast. The toast is fine. The regular toaster also made the toast fine. So that was nixed. I don’t need Fable for that Same deal with the content repurposing. So turning a podcast transcript into post ideas, pulling quotes, drafting captions, that mid-tier model has been great at that for at least a year that I’ve been using it.

[00:09:30] And if that’s most of what you want to use AI for, that’s a lot of therapists, that’s pretty much what you’re using for. You don’t need to pay for the top model, even the Opus model, you don’t need. You can go down to Sonnet or Haiku and have a lot of success with just using those models. They’re much, much cheaper.

[00:09:47] Now, the other task, this is where it got a little bit interesting. The fact check I told you about earlier, that was Fable, and I could feel the difference with that because catching that claim is half true. It’s a different kind of task than writing words. So it had to notice that free technically meant included in paid plans, so it picked that up right away.

[00:10:06] And then there was a usage cat buried in the fine print. And Anthropic’s own statement about bringing Fable back contradicted the entire last chance framing. So it went out and searched the internet, found all that information, brought it back and compiled it for me, which I thought was really, really good.

[00:10:21] Better than like Perplexity’s deep research or ChatGPT’s deep research that I’ve used before. So finding that thing that’s slightly wrong and knowing why it matters, that’s where the smarter model really is earning its keep there. So if you’re doing like deep research or deep analytical thinking, then the model might be the best choice when it comes back out and is available to everybody.

[00:10:41] And then the podcast synthesis, I asked it to look at my podcast, all 100 plus episodes, plus what’s trending in AI news this week, plus what I’ve personally been working on inside of my Claude and my computer in general. And then I told it to pitch me this week’s episode, and it didn’t just give me generic topic ideas.

[00:10:59] It noticed that episodes 101, 103, 104, and 106 were all like secretly circling that same question, what’s actually worth paying for in AI? And it connected that thread into the Fable news. And I hadn’t really seen that pattern in my own catalog. It found the through line in my body of work faster than I would have, or maybe I never would’ve thought to like connect all of that And that one part gave me chills a little bit.

[00:11:25] It was like, whoa. It just, like, looked at all these episodes, looked at everything I’ve done on my computer and in my cloud, came back, compiled it, and, like, gave me this really comprehensive idea based on what’s happening right now with AI and also what it’s seeing me do with my work. And it pulled it all together for me.

[00:11:42] In the offer pressure test, that was the same story. I gave it a course idea I’ve been noodling with on the niche, the price point, and asked it to poke holes in it. And it asked me the kind of questions that my old business coach used to charge $300 an hour to ask me. Like, “How is this different than your existing memberships?”

[00:12:00] And, “What happens to this offer if the tech it teaches changes in six months?” Which I’ve had happen to me quite a few times with AI. Questions where I was like, “Oh, yeah. Okay. Yeah, I need to sit with that. That’s a really good point,” or, “I need to rework this or rethink it.” So the strategic thinking was well worth using the Fable model, in my opinion.

[00:12:19] And then the paperwork audit. I pasted my consent form templates and asked what I was missing or what was unclear. This is for my coaching clients, by the way, not my therapy clients. And it flagged gaps that maybe a cheaper model would have skated right past or just agreed with. You know, a lot of the models just they placate you, right?

[00:12:39] So they just say, “I think this is great, Kim. Maybe change this word or that word.” But it, like, really broke it down for me. So when I did that paperwork audit, those were like my blank templates, my documents about my coaching practice, my products, consent forms, no client names, no client information, nothing like that ’cause they’re coaching clients.

[00:12:59] Regular Claude is not HIPAA compliant, and nothing with client data will ever go into Claude, so just wanna remind you guys of that. Your clinical notes stay in your HIPAA compliant tool built for that, which for me is Berries. It has been for a long time with the BAA in place, business documents in the everyday AI clinical work in the compliant tool.

[00:13:22] Make sure you don’t ever mix those streams. I know you guys have heard me say that a million times. So lying there in bed looking at my five little experiments, here’s the pattern I found, and I wanna give you the language for it because I think it makes every future model decision much easier for therapists that are already strapped for cash, you know, aren’t sure they should be paying for a higher model.

[00:13:41] So there are word task and then there’s, like, plug the holes task. A word task is where you’re asking the AI to produce writing in a known shape, post captions, emails, handouts, show notes. The path is pretty clear. The voice is trained if you’ve done that already. The AI just has to go and execute what you’re asking it to do.

[00:13:59] And then mid-tier models are fine for those, and they have been for a long time, as long as you’ve trained it properly. A holes task is where you’re asking the AI to find what you can’t see. So poke holes in my offer. Tell me what’s missing from my paperwork. Tell me which part of this viral claim is not true.

[00:14:17] Find the pattern across hundreds of podcast episodes. So there’s no template for that. It’s judgment, and judgment is exactly what these frontier models are getting better and better at. And what I’m reading out there is that really if you’re a coder, Fable 5 is amazing, but that is not the kind of stuff I’m doing really.

[00:14:36] So, like developing new games or products, that kind of thing that need, like, coder level mind, I’m not doing that. That’s the code side of Claude. I’m using, like, the co-work side for a lot of what I’m doing. So you can still use it to plug the holes, so to speak, but you don’t need it for a lot of what most of us are doing as therapists that are not coders.

[00:14:57] So when a new smartest model ever drops, and there will be another one I’m sure, don’t ask, “Should I use the new model?” Ask, “Is this a words task or a holes task?” Words task, use the cheaper model and keep your money. Holes task, that’s where the expensive brain might be worth spending a little extra money.

[00:15:15] So that’s it. That’s the whole decision. And people ask me every single day, “I don’t know what I should be using. Should I be using Claude? Manus Perplexity?” Should I be using, you know, Opus, Fable, Haiku, Sonnet? Which one should I be using? So these are some guidelines to help you decide that a little bit better.

[00:15:31] So three things I want you to take from today. First, fact check the gurus before you repost them or share these viral posts. Make sure it’s right. The free Fable claim was half wrong, and that urgency was fake, and it took two minutes to find that out. So make sure you check that before you, like, fall down the like, “Oh my God, I gotta do this right now” train.

[00:15:49] Make your AI show you if it’s true or not true. Second, word tasks versus whole tasks. If you’re mostly generating content, the model you already have is almost certainly enough. Save the top-tier model and the top-tier price for the really complicated work that you wanna do, like building out a new product or building a super sophisticated app that you’re gonna try to sell, that kind of thing.

[00:16:11] Third, and this is the one from someone that’s basically been in bed for a week and a half, build your systems before you need them. I’m watching my business basically run itself this week while I heal, and past Kim is the only reason present Kim gets to rest a little bit more this week. So whatever version of this you’ve been putting off, the automations, the team handoffs, the documented processes, start it now while you don’t need it yet.

[00:16:35] If you wanna see the actual tests and the prompts that I used and what Fable said versus what the cheaper model said side by side, I’m gonna be showing all this to the members inside the Clinical AI Club and Thera AI Hub this month. We have three meetings this month. I actually added a new monthly meeting, which is just called Coworking, where I’m just there to help if you’re stuck with any of the content that I put into the memberships that month.

[00:16:59] Three meetings, two masterminds, and one coworking meeting every single month where we talk about all of this in depth. All right. Thanks for hanging out with me from my recovery room. Be nice to your future self this week. And if you have time, set up some automations that are gonna run for you after today.