Ep 298: Claude Skill: Research to Possible Solutions

I'll introduce you to my Claude Skills, which are available for purchase at https://deltacx.academy  Currently, I have one published, so... let's learn more about what it can do!

Full web page on the skill here: https://atomicpmf.com/claude-research-skill

Research to Possible Solutions (released in early April 2026) covers research analysis, synthesis, user segmentation, problem statements, root causes, strategies, and exploring possible solutions.

Transcript

SPEAKER_00 0:55

Welcome, low ego action heroes and phoenixes. I'm Debbie Levitch from Delta CX, and welcome to episode 298, where I'm going to talk about my Claude skill for research, going from transcripts and data through analysis and synthesis and into a little bit of problem solving, if you would like.to. Sorry, DCX.to that will dump you right on our links page, and you can find everything that we've got from my books to the free Discord community, the free Patreon, all kinds of coaching and even more stuff coming. So, um, what's this Claude Skill all about? So I built the Claude Skill for my own use because I've been using Claude in uh my research work for uh gosh, over a year and a half now. And so by from when I'm talking to you, which is uh late May 2026. So I built the Claude Skill for myself, and then I decided I would make it available to other people at a low price. Anna Lucia says I'm using it right now. Thank you. Yep, she got an early uh version of it. Well, it's still the same version, but she's she saw it first. Um, and so if you are curious or interested, you can check it out at atomic pmf for product market fit, pmf.com slash clawed dash research dash skill. It's also linked from the Delta CX website on the training and courses page, but that'll take you right to it. And it's also on the DeltaCX.academy uh gumroad uh store where you will buy it. It is currently 50 American dollars as a one-time fee. Um, and I also wanted to build the skill because I wanted to make sure to build something that helped me do my work faster without sacrificing quality, and I wanted a skill that uh wouldn't replace me, that it still required my abilities, my critical thinking, my research sessions, uh, since I'm working from transcripts of real interview sessions. Anna Lucia says, Hey, important thing, Sonnet 4.5 was officially removed from free users. I was able to change to 4.6 without having to start a new conversation. Yes, they are phasing out Sonnet 4.5. The example I'm gonna show you live today, I am in Sonnet 4.6, but of course, models will change over time. But I do want to mention that I believe Sonnet is the best model for this skill. I think Opus is overkill and going to burn tokens way too fast. And um hi, person in the Netherlands. Um, we don't speak Dutch here. Um uh and I think Haiku is not enough. We tested Haiku and it just uh couldn't handle the deeper critical thinking and a lot of the data. So uh if you're using Claude, Sonnet is going to be the best model. And just okay. Um, and so uh what we're gonna do today is I'm gonna show you live some of the Claude skill. I only have about 45 minutes today, since after this stream, I do another stream on my other channel, Vocal Coach Debbie. And today we're going to talk about better breathing techniques for singing. So you can join in on that as well. Um, so I am going to share my screen, and because I knew we would only have 45 minutes, I started the skill earlier today, and so I'll show you how far I got, and then we're gonna do some things live together. Um, Anna Lucia says, I was afraid my work would go out the window at the end of Sonnet 5. No. Um, okay, so sharing my screen. So note, um uh it's my actual skill, and I'm using nine transcripts from a real research project that I did in February. Um, I've anonymized them. They all related to a project where I interviewed uh people who had bought real estate in a country that was not their original home country, and so they were all different people from all different countries buying in different countries, and this was going to help my client, who was a government, uh, looking at uh relaxing the rules for foreign buyers, and they wanted to understand foreign buyers' experiences so that they could make buying a better experience. So that gives you a little context for the project. Oh, good. Anna Lucia says I've got some notes to help you improve the skill. Super, definitely send those over. And uh, if you do buy the skill, you will get free updates and upgrades to it. So definitely head on over to DeltaCX.academy and grab yourself a Claude skill. More skills will be coming, hopefully, over the summer. Okay, so um flipping over to my screen. Basically, once you tell Claude that uh you want to start the skill, and by the way, you just install it uh under customize, you upload the skill, and it's got sub-skills. And yeah, the real estate project is back. I've only got three. So by the way, we did test this with a set of five um transcripts from interviews from one project. This is a nine interview project. I've done it with a a seven uh twice, I think. Uh two different projects that had seven people, and then we also tested it on a project that had 28 one-hour long transcripts, and it it took a long time for it to get through those, about an hour, but it did a really great job. So um, so I've tested my skill pretty rigorously, but I'm always happy to keep improving it. And remember, you'll get those upgrades too. So the skill has sub-skills, which uh some are optional. For example, if you don't want to look at personas and segmentation and customer profiles, you can skip that one. Uh, research analysis, you're gonna kind of have to do. Strategy is technically optional, but I think it's really important to report strategy and deliver more things that can help our teams make better decisions. And the strategy uses the strategy template from my new Atomic Product Market Fit book, which has blue on its cover, so it's being blue screened out. Um, but this uh skill is kind of a cousin to that, and then possible solutions, and again, depending upon your job, you might say, Well, I'm not allowed to suggest any possible solutions, so I'm not gonna use that. Or, hey, this is for my startup. I do want to play with possible solutions, so these are optional for you. So, this is how much chatting I did with Claude today. I definitely spent a lot of time uh on this to get, and then I was just pinging it to keep this uh instance alive. So it's already created two files. Let me show you the first one that it creates. This is called participant, and basically what Claude does is as we uh start looking through the project and at the transcripts, it um uh it will make notes. So this is a little bit out of order, things happen in different order here. So it will actually make notes as it reads the transcripts as if it were your note taker, and so that way, especially if you have a lot of transcripts and it can't hold those all in memory, of course, it can work from the notes. So really key. Um, can you upload your own notes? You can, but I found that Claude sometimes got confused between what were things that participants actually said and what were things from your notes. So if your transcripts are are good, then work with those. You can upload your notes if you would like. Um, or as you go, you can just tell Claude things it didn't know because they were things you happen to observe or things it won't catch in the transcript. So you can see it takes some extensive notes about each participant, and there are nine here. So that way, again, if you uh lose this particular instance, you come back a day later, it's gotta read stuff, the lunches, yes. Um, and then here we've got a section where we talk about who are the actors and who are the missing actors. And I think that's an interesting thing to add to our reports in that here we are talking about buying property, looking for property, buying property after we've bought the property. And there's so many people involved in that because it's really a service design universe. There's so many people in our experience ecosystem that we didn't talk to. We only talked to 10 people. I spoke to 10, but I didn't upload the 10th one because their um English wasn't very good and it was a bit of a tough interview. So nine strong interviews. Um, they mentioned all kinds of people who we didn't meet. So remember, we don't have the perspectives of real estate agents, notaries, lawyers, banks and lenders, tax advisors, and and other people who are involved. And that's interesting when you want to remind your team that there is room for more excellent research later to get more opinions and perspectives and experiences from other people in the experience ecosystem. Um, other uh actors as well. People had to involve their consulate or a land registry or a permitting department, um, friends and family, we didn't talk to them. So there's also parts of the uh skill that ask you about the scope of the project, uh, what should be anonymous. So participants are anonymous, but any companies or systems or uh things they name will not be anonymous, but any people they happen to mention by name, we will keep anonymous. Um, the uh skill will also take a look at the quality of your data. So if you're trying to do a really good analysis and synthesis and possible solutions and problem statements and root causes, and you just have a survey, the system's probably going to tell you this isn't the right set of data for for this. So there's some quality checks built in. Um we also uh broke people into groups. So we had segments uh one, two, and three based on their knowledge going into the property shopping and buying process. Some people had none, some people had a little, and some people had bought once before or multiple times before and really functioned as an expert. So we kind of have this along, we have these segments or personas along this knowledge continuum. Um, that was something I spent a good amount of time on uh today with Claude. Um, and then the second document that it creates, it will create three documents in this, and that's just to break things up so it's easier for you to find pieces later when you're making your reports. The second document is analysis. We just started this one, so it's much shorter. Um, it repeats some of the information from the other document. So here's our segments, and we also have the actors, and I'm not seeing the um, I'm gonna ask it to add. We did go through topics, and I know that it, I think it added them to the other document. Let's check. Oh, wrong one. Participants and what topics is about is topics will take a look for um let's see, topics, topics, topics, is it in the no, it's not in the document, so I may need to ask it to add that. Um uh, I think segment three. Yeah, so we'll have to figure out where the topics come in. But basically the idea is that hypothetically, the task or process or thing that you're studying might really be more tasks or topics that uh that should be separated out. So as you can see here, we taught Claude said, Okay, I think I see the buying process and I see the post-purchase. Can we do two topics? And I said, actually, I think it's before the purchase, the actual purchase process, and then after the process. And Claude said, Oh yeah, okay, I like that. Let's use these topics. So that means we're not just going to look at all of the data in one bucket, we're going to remember that the data happens across three uh tasks, and you could have even more subtasks there if you wanted to. So we are also going to be thinking about things in terms of tasks. So um, I did also tell Claude that this is a demonstration that I'm doing live. So it knows that this is part of a demonstration, but I told it to treat the skill as completely realistic. And if anybody has any questions while you're watching live or you're seeing the archive on YouTube later, you know, give us a thumbs up, hype it, and uh give us a comment. I'm happy to answer any questions about this skill because I want it to help you. I want it to make you faster and better while maintaining quality and without teaching people you can be replaced. So uh we are live. Uh let's pick up where we were, which I think was problems. So Claude was starting to list for me and catalog the main problems that it saw people having uh in the uh, let's see, we're at the right moment. I have 11 problems across three topics to write into the analysis document. I can read them to you before the break. Want me to write them in and we can review them on stream, or do you want to walk through them conversationally first? So let's roll back and I'll show you the problems it came up with. And it uh broke them down by our three topics. So it said for topic one, we've got problem, so it's calling it T1, and then problem one. No accessible authoritative information about what buying in a foreign country actually requires. So nobody gets checklists, nobody gets guidance, nobody is told pretty much anything. You have to do a lot of research yourself, you have to know what questions to ask, and of course, you don't even know what you don't even know. Uh, which is which leads to problem two. They don't know what questions to ask. Uh some people named this explicitly. Some people bumped into properties that were illegal that they didn't even know to even think and imagine that the property would be illegal. Uh, finding a trustworthy, competent agent is a lottery with no safety net. Uh problem four, professions in buyers' home countries have no, oh sorry, professionals in buyers' home countries have no cross-border expertise. So we found in general the uh the people they were used to working with at home had no idea how to buy in another country. Uh Analusia has a question. A question about the skill itself. Did you add anything about potential research points for the future if you have to continue within the same theme or something similar? I think the potential research points for the future would probably come from the agents we didn't talk to or the roles and people and actors that we didn't talk to. Um, but I didn't add anything yet. I haven't changed the skill since you saw it uh in early April, I think it was, or mid-April. So I've not added anything. So if you have feedback around that, please let me know. Uh, problem number five, tax and legal complexity is hidden until it's too late to factor into the decision. So again, a lot of things people didn't know and didn't know to ask happening across topic one, that pre-purchase experience. Um, so you can see just as an example, because we've got only 30 minutes left and I want to blow through more of the skill, we've got some problems. So uh I'm just gonna pretend that these are good and tell Claude um uh these are good. Let's document them and continue moving through the skill. And so what it should do is it should add these to the uh the document that we have. And of course, at any time you can download the document, you can save backups of it. These are good ideas. Um, so you can see right now it's doing what's called a string replace, where it is adding some lines, removing some lines, and um, and eventually you're gonna see this uh spin. There it is, it's spinning, and that means it's refreshing this document. And now if I eventually scroll, uh, we should be able to find what it just added. Let's see problems identified. Let's see where did it add them? Atomic analysis too, okay. Ah, here they are. I had to refresh it. Here we go. So it added this, and again, it's a working list. Don't show this necessarily to your client. We're gonna now find root causes. But of course, uh one thing we're gonna do later in the skill is we're going to write problem statements. So right now we're just trying to catalog some of the key problems that we heard during this research. Now, there are times where I might have fewer than 10 problems, and there are projects I've done where there are more than 30 problems. So remember that a long list of problems doesn't necessarily mean that Claude is wrong or that this was done badly. You can always tell Claude, hey, I think problem one and problem two are really the same. Let's combine them. And so everything in the skill, every moment is an opportunity to uh communicate with Claude and to collaborate and to be critical thinking partners. Claude will challenge you, you will challenge Claude. So now it's going to try to find what the root causes are based on the research conversation transcripts that it has. Root cause one, there's no information infrastructure designed for foreign buyers. This is underneath a bunch of the problems. The process knowledge that determines success or failure lives in informal networks, family members helping out, friends who happen to be lawyers, agents who happen to speak your language. When these exist, people succeed, and when they don't, they're they now here we have flying blind, which to me is ableist. When I do see Claude using ableist uh language or any sort of bias, I do tell it to fix that. Uh, or they don't buy it all. Root cause to professional ecosystem was built for domestic buyers and not redesigned for foreign ones. Uh yes, so so far, as someone who ran the research and had all of these conversations, these sound very true to me. Uh Anna Lucia says, good ideas and the only way to cover your butt if Claude is having issues. Yes, I think that she's talking about when I said, Don't forget to download some backups of these documents. Um, no one owns the buyer's journey end to end, everybody's handling it in pieces. Root cause four, the gap between foreign buyer status and the formal system creates friction at every touch point. Now, I think in the end, there are probably more root causes than these four, but I would say that for the purposes of just trying to blow through this skill and show you what it's capable of, I'm gonna say yes to a lot of things that Claude is doing, and I am not gonna challenge it on as much as I normally would. Normally I spend a half day, sometimes a day, sometimes more with this skill, but that has reduced my time with analysis and synthesis and reporting from sometimes weeks down to sometimes a day or a couple of days. So I'm very excited about my own skill. So I would say um, for our purposes, let's say these root causes are good, but in reality, I would spend more time with it. I would read them out loud, I would double check it against my own memory and knowledge of the research sessions. I would ask it, hey, what about this other thing? And you know, we would go back and forth until we had uh a set that we liked. Um I'm uh so I'll say I'm okay with these, but uh please don't use able list terms like flying blind. Please replace that with another phrase, and we can catalog these in the MD file. Thanks. Uh Claude says, I'm going to say navigating without a map or left to figure it out alone. I think that's good. On the question about root cause three, I'll keep it separate, even though it is partially a downstream of root cause one and two. So that's interesting that root cause three is its own root cause, but also semi-caused by the other stuff. So that's no one owns the buyer's journey. Again, you could always go back to Claude and say, I don't want downstream stuff in here. I only want the root, root causes. Up to you. Or you might include some downstream things. So it's really up to you and what you can report on for your team or your client. So that means that somewhere in this document, we now, here we go, we have our root cause candidates. So again, these are all things that you can include in your report if you would like, or you can just keep them as part of your process with Claude to eventually get to the stuff that you will put in your report. Up to you. So now Claude says, um, problem statements. These are client-facing versions, crisp, grounded in the data, written with gusto. That's not in my skill, but that's very cute. I'll draft them and we'll review them before they go in. Give me a moment. So PS stands for problem statement. Um, unfortunately, it later also stands for possible solutions. So that's something I'll eventually fix where these have different abbreviations, but for now we can tell the difference between a possible possible solution and a problem statement, especially since uh problem statements I've given it a sample. So it's got a uh uh template for problem statements in it inside the skill. It's the same template I use in the Atomic Product Market Fit book. And um we're looking for the problem statement to give us a little bit of the who, what, where, when, how, and why, but to also show the impact it's having on our company or uh your client because sometimes people don't care about customers having problems until they understand the impact on the company itself. So I'm gonna be checking that Claude has written a good customer-focused problem statement, not a problem statement from the perspective of the company, that also includes the impact on the company. So let's take a look. The information gap is structural, not personal. Foreign buyers cannot succeed at cross-border property purchase without access to insider knowledge, and the system provides none. The processes, the risks, the required professionals, the taxes, and the timelines are not documented anywhere accessible to a foreign buyer starting from zero. Every participant who succeeded did so because someone they knew had done it before. Every participant who got hurt did so because that person wasn't there at a critical moment. This is not a buyer education problem, it's a system design problem. So I'm gonna tell Claude to please rewrite these problem statements using the template that's in the skill because I think it just went for problem statements. Um can you please rewrite these using the skill? Um, please make sure you're following each step of the skill as we go. Um, these problem statements don't seem to match my template and parameters. Thanks. And to that, Claude says, You're right. Let me go back and read the skills problem statement. Now, this is where some people get angry and freak out and take a screenshot and write a LinkedIn post. Can you believe that? It didn't follow my skill, and then when I pointed it out, it's like, you're right, and it's sorry. Hey, you know what? Calm down. This is the way that AI works right now in 2026. It misses things, it improvises, it hallucinates. Get with the program, no reason to freak out, no reason to cause yourself more stress or negativity. This is just the the game. And that's why there are times where Claude is gonna challenge you in this skill if you're using it well, and there are times that you're gonna challenge it. Uh, I can see what I did wrong. My problem statements were essays, not the tight specific business cost grounded skills. Um, and I skipped step 12 and step 13, refining root causes together. Let me back up properly. Um, so there you go. So you can always say to Claude, hey, can you double check the skill? Did we skip any steps? You don't even have to know what the steps are. You can just say, hmm, you know, just checking in, did we skip any steps? Are you following the skill step by step? So you're always welcome to ask that. I know what comes in what order in the skill. You don't, that's fine. You don't have to learn it. Knowledge design. You can just ask Claude, hey, just to check in, are you following the skill steps in order and moving through all of them? So um, so notice that step 13 says, we have our root causes, but what doesn't work with them? What's missing? What do you know about this client or ecosystem that wouldn't be in the data? What would you combine or split? Let's challenge each other. So again, this is where I might spend 10 minutes, a half hour, hours refining this. But for the purpose of trying to blow through the skill and show you what it's capable of, we're going to say the root causes are pretty good and it's okay to move forward with them. Uh thanks for stepping back. I think the RCs are good and we can move to the next step. So now what it's doing is it's running its own check. It's checking if every problem that it documented before, which were those 11 problems across the three topics, if if there is at least one root cause that can explain that problem. And if not, then we must be missing a root cause. So Claude says, hey, there is an orphan. So I didn't check it because I'm blowing through it, but it has a step for checking. Uh topic three, problem two, post-purchase property management and maintenance. This is real data like participant five's renovation ordeal, rural contractor issues, but it doesn't trace to any of our root causes. It's more about local market infrastructure gaps than the foreign buyer information system. It could be outside of the client's direct control. What's your instinct? Add a root cause, note it as outside the client's control, or merge it somewhere. Well, how could what could be the root cause of post-purchase property management and maintenance sucking so bad? It could relate to the root causes we've already identified that there aren't uh centralized systems or a centralized database or an easy way to find these particular vendors. So um I'll say perhaps the root cause relates to uh no centralized system or database for vendors that might cater to uh people who don't speak the language. Thoughts? Improvement? I usually bounce it back to Claude and say, now you think about this, make it better. Don't just go with what I've said. Um, Claude says that's the data supports that. Is it a fifth root cause, or is it root cause one in disguise? Nice critical thinking. Root cause one was there's no information infrastructure designed for foreign buyers. And that could end up being lawyers, real estate agents, property managers, repair people. Yeah, I think Claude is right. So Claude is saying, hey, we can either expand root cause one to include this, or we can add it as a separate root cause. I like the critical thinking going on here. Um, yeah, I like this. Let's um build RC1 out further to indicate people like vendors and property managers and repair people. I'm typing terribly, but I'm just going with it today. So Claude's now going to update our root cause, which is over here. Um, in our document. So again, the great thing about an MD document where we use string replace is it's always being updated, and you don't have a situation where Claude is always trying to write the full thing out in chat and you're trying to reread the whole thing. I'm just uh taking a drink of ice water because it got it's suddenly pre-summer over here. Okay, Claude says now everything okay, now you can see this is longer and it includes that it covers both uh uh topic three, problems two and three. So now Claude is going to draft a problem statement based on my template. The first thing that I notice is it's way too long. My uh mine says uh try to stay around a hundred words or fewer. I feel like no matter what this says, it's too long. So I'm going to say, before I even read this, before I even read this, holy cats, it's long. Can it please be more streamlined? Because I find that we do want to present problem statements in our reports or slides, but nobody wants to read something that long. So let's try. So Claude says closer to the e-bike example, and you're thinking, e-bikes, that's built into the skill as a sample statement, and it also comes from my book. So let's read this and see what we think of it as a problem statement that we expect to really help our teammates and clients uh understand what's going wrong here and why it's important that we fix it. This says foreign property buyers have no formal system to turn to when figuring out what buying in another country actually requires. None of it is documented anywhere, accessible to someone starting from zero. Every participant who succeeded did so because a personal contact had already done it. A family member's warning about illegal properties, a friend lawyer, a same language agent. Uh, Analusias says, Thanks for the tip about the templates and parameters. I didn't know it had one, and Claude went wild with my experience. Yes, there there are lots of templates and samples built into the skill. So you can always, uh if you're not sure, you can always say, Hey, are you working from a template or a sample? And if you don't like my template or sample, give Claude yours. Hey, I don't like the way those problem statements look. This is how I want to write them, and Claude will say, okay. Um, so continuing the problem statement, when that contact wasn't there at a critical moment, buyers discovered what they'd missed months later. Unexpected utility bills, surprise maintenance fees, obligations nobody explained. The gap doesn't close after purchase either. Foreign owners have no accessible way to find local vendors, contractors, or property managers without language fluency or an in-country contact. One participant spent their entire first holiday summer managing a renovation or a deal because no formal support system existed. For a government seeking to attract foreign property investment, this is a direct constraint on market participation. Buyers without the right personal network don't buy or buy once and don't return. Markets with clear, accessible infrastructure for foreign buyers are winning those transactions instead. So then I would ask myself, is this problem statement enough where a client, a strategist, a product manager, a UX designer, an engineer, a marketer, would they really get what's going on here? The short version. We can always give them more data, and we're certainly going to give them more slides or more of a report than just this. But is this a good short version? Should it be shorter? Should it be longer? Should it have more information about how this can really hurt our client or our teammates? Totally up to you. That's why at every step of this skill, it is a give and take. It is a critical thinking partnership. So Claude's asking me, is that better? Yes, I think that is better. And again, normally I would spend a lot more time rewriting this. Sometimes I copy and paste it into notepad, I rewrite it myself, and I pass it back to Claude. And I go, what do you think of my version of this? Let's make my version better. Let's keep going with it. So again, don't be afraid to be a very active part of this skill. This is not here to do your work for you, it's here to do the work with you. And you are welcome to take it over at any time and pass it back. So I'll I'll imagine for now this is fine. Uh, yes, let's please uh document this one. So uh it should be updating the file because I have the skill adding one um problem statement at a time. Why? Because you're then given the choice to write more problem statements. You can write all the problem statements while you're at it, or you can move into well, what might be a strategy to solve that problem and what might be possible solutions for the problem? Those are additional skills that this handles. And so you normally I would just keep writing more problem statements so that I have all of my problem statements documented and check if they overlap, check if they work together, check if I'm missing anything. But for now, because we only have another 12 minutes, I'm gonna blow through other aspects of the skill. Um, but hopefully you're getting some good ideas about what this is capable of and how it might help you in your research-related work. Um for now, let's move on. We can always circle back to other problem statements later. Um, what's next? I think it's strategies. Um yes, let's uh skip four now. Just uh since I'm short on time, it's cute that it's trying to get me to continue working on this because I normally would. Uh let's uh see more of the skill. Okay, so the next thing that happened is Claude creates a third document. This document is called synthesis, and it's going to uh be where a lot of the uh next pieces of work live. Umtice it's giving you a short version of the root causes to refer back to. But remember, we can always close this file, open up our files, and see the three documents that we have made in this chat. Again, you can always download them or save backups of them, and you can always pull pieces out of any of them for your reporting. You can also break out of the skill. You can say, hey, wait a minute, Claude. I need to uh do a write-up on each participant. I need a bulleted list with all nine participants and give me one sentence on each with what country they were from, which country they were buying in, and whether or not they paid cash. Claude will make you that and then say, hey, let's add that to the participants file. Just you can break in and you can have it do additional things. So just remember that. So now we're in the synthesis file, which is where we're gonna catalog our strategies, which of course is optional, but I recommend you at least give it a try. And we're gonna document possible solutions, which again is optional, but I recommend you give it a try. So Claude is using my strategy template from uh one of the later chapters of Atomic Product Market Fit. You may like this template, you may not like the template, you may have your own template, you might want to change it. Totally up to you. But we start with a core strategy. We have our problem statement repeated here. So again, this is really designed to copy and paste right into a report or a slide deck. We've got our root causes, so we can go back and copy and paste those. We've got strategic North Stars, goals and how we'll win, possible KPIs that our client or our company or team could measure to see if we're on the right track, and risks. So that's my strategy template, and we'll take a look at this one. Um, we've only got about 10 minutes left together. Hello. Uh, and so uh core strategy governments that want to attract foreign property investment don't just change the rules, they design the experience of buying, making it possible for any motivated buyer, regardless of their personal network or prior knowledge, to navigate the process with confidence and finish as a satisfied owner. So I would probably have it take out the long dashes and un-AI this a bit, either by rewriting it myself or suggesting to Claude a rewrite. Um, is that a great core strategy? That that's okay. I think for my taste, it's a little softly written. I'd probably want to write something that's a little bit uh stronger and a little bit more of a guiding light, like, hey, this is your core strategy. But remember, your core strategy is to make buyers successful and make the system as easy and transparent for them as possible. So I might say, um uh let's add, let's adjust the core strategy to indicate that the uh foreign buyer program is only successful when buyers are satisfied and successful. And um, I forgot what I said after that, so we'll just leave it there. But normally I wouldn't be talking, I would just be typing this is our same problem statement, this is our same root cause. Let's take a look at these strategic North Stars. And the idea is that the North Stars will guide your team or your client or whoever to solve this problem with these guidelines in mind. So they might be ethical, they might be legal, uh, or they might just help people stay focused on the original problem uh so that they don't design something that works against the problem or doesn't really solve it. No insider required. A foreign buyer with no prior personal network in the target country should be able to navigate the full purchase process and post-purchase ownership without needing to know the right person. Access to information, professionals, and local services should be a system feature, not a privilege of connection. Every information gap currently filled by informal networks is a design opportunity. I like that. Um, so this is an that's an example of strategic North Stars. Claude's written three. If you want one, that's fine. If you want ten, that's fine. You work with Claude to update and change this. Uh, ownership doesn't end at closing. Very true. Goals and how we'll win. Foreign buyers can understand what buying in the target country requires before they commit, including costs, taxes, legal steps, and required professionals. Uh, yes, so again, we've got bullet points here with goals and how we'll win. This is your opportunity to rework this with Claude. You want it longer, you want it shorter, you want more of something, less of something. You want to add something, you want to remove something. Work with Claude on making it what's going to be great and compelling in your report or slides or whatever you're delivering after your research project. For me, it's usually slides. So I'm usually taking uh bits of this or all of this and pasting it onto one or two slides and telling people this is our strategy related to our uh root cause one, solving that root cause at its root. We've got our KPIs, which are some things we would expect to increase or decrease based on solving this well. Um, Claude also went wild on me in this part. I'm making it check the templates. Yes, there is a template here. Um, and so if it's doing some weird things, you have the opportunity. And remember, you always have the opportunity to say, hey, please double check the skill. Are you following the steps and any templates or or samples you have in the skill? So for example, a uh and Claude just uses letters here, so you can fill in later with what percent increase you think it should be, or maybe someone else on the team or at the client will put in their number of what they think the increase or decrease will be. But um, but they but it's there. So there should be some sort of increase in foreign property transactions completed without reported failures, some sort of reduction in time from decision to closing, some sort of increase in foreign buyers who report feeling informed through the process. So things, some of these things could be based on interviews or surveys or other metrics. And finally, what are the risks? And in my uh template, Claude is looking at risks in five areas: market risks, technical and feasibility risks, resource risks, competitive risks, and CX risks. And sometimes I find that Claude underestimates some of the risks. So this is where you might want to look at it and say, well, actually, there is a CX risk here. You're saying it's low because good information doesn't create new friction, but I think the risk is really medium because blah blah blah. Or I think the risk is really high because blah blah blah. So you have the opportunity to work on this with Claude. You never have to accept the first thing the skill gives you. You can always make changes and update. So I've asked Claude to rewrite the core strategy a little bit, and it says it is going to update that now. And then I assume once it's done, we're gonna see that strategy showing up in our document. Um, so this should refresh soon and pop up. And then Claude says, Normally I would ask you if you'd like to write strategies for the other um uh problem statements, but we only wrote one problem statement. So would you like to look at possible solutions? Um, and that is a separate subskill. So, okay, so here's our strategy, and now we've got possible solutions. And um, I'll just show you the beginning of this because we won't really have time to get into it. We will have to wrap up today because I am doing another stream right after this. So that's gonna be my my Tuesdays. Um uh we can start the possible solutions skill. Um, so it is a it's technically a sub-skill uh of the same skill. So when you get your skill, a skill in Claude is technically a zip file with multiple documents, so you can have a skill with subskills. So just so you know, there are three ways that you can brainstorm possible solutions if you'd like. Um, Claude will suggest an idea and you'll go back and forth on it and challenge each other. Um, you can bring an idea and go back and forth with Claude and challenge each other, or Claude will ask probing questions to help you, or maybe you and other people, brainstorm possible solutions. So again, that way it's not just, hey Claude, come up with ideas. Okay, let's write them down. No, we're going to challenge them, we're gonna work on this together, and then there's room for you to bring ideas as well. Anna Lucia says some of the KPIs Claude suggests are not okay either. We have to check everything. I remember one that I had to tell it that had to be self-reported. There was no possible way to get that information without asking people, and Claude was like, oh yeah, I'll fix it. So exactly. It might have suggested I didn't look at all of those KPIs. There might be KPIs there that don't make sense, that are too hard to measure, that couldn't be uh in your data analytics or surveyed or anything. You just can't know. So that's where you have to read through this. Don't just copy and paste and give it to your teammates. Take the time to stop and read through it. I'm only not doing that because this is a sample demonstration. But when I actually use this on my project, as I said, I spend a half a day or a full day on this skill. And then from there, I'm copying and pasting and usually rewriting and rethinking some pieces as I put them into slides. Um, so let's wrap up for today. That is the one-hour version of the Claude Skill. It is available for purchase uh on the Delta CX.academy store. Um, if you buy it on Gum Road, I believe if you are in a purchasing power, um, purchasing parity power, oh gosh, I always forget what three P stand for. But I know it is giving people some discounts uh if you are in certain countries. Um, so otherwise, it's only $50. Please pay for it. It's going to save you way more than that in time. Um, it I I do tend to use it with a Claude Pro account. If you have a free account, uh, from what I'm hearing, especially from Anna Lucia, you're barely going to be able to get through the skill. It is just kind of a little bit uh memory hungry, especially these earlier parts where it's reading all of the transcripts. And um, that was my reminder to end the stream. Um, so um, probably better with Claude Pro. Can you use this with ChatGPT or something that isn't Claude? Yes, you have access. Once you buy the skill, you can open up the skill and read all of the files. You can take the files and put them into Chat GPT, Gemini, anything else as prompts. And you can just say, hey, I'm gonna be working from a skill. I can't upload this skill to you, but I can paste pieces to you as as you go. Uh Anna Lucia says it's taking me a lot of time to do this with my free account. Yeah, and usually when I teach this skill to people, it's uh to get in really much deeper than we have today usually takes me three or four hours. Um, and for that, I typically buy Claude Max just so I don't run into the five-hour window uh where I'm out of uh tokens because I've only got three or four hours to teach this to you, and I don't want to run out. But outside of that, I use Claude Pro. And if I run into the five-hour window, I take a break. Um, but I usually don't, it's it's not heavy enough to burn um a week's worth of tokens uh because it's just all chat. There's no there's no images, there's no sound files, and so um, yeah. So warning this skill, I'm really proud of it. I'm endlessly excited about it. I think you're gonna love it, but um much better with paid accounts. Um oh, thank you to the new Twitch follower. Um, whether uh again, whether that's paid chat GPT and you want to paste this in, because that's how I tested this skill originally. I just took pieces of the skill and I pasted them into a chat. Like I would say, okay, here's all of the research prep skill. Let's go through this step by step. And then I would say, okay, here's all of the segmentation skill. Let's go through this step by step. So you can always um paste these sub skills into a different AI system and and see how it goes. Um, we also had someone who came to one of our courses who um, okay, so this is uh spam from LinkedIn. That's just embarrassing. That's that's just you should you should be ashamed. Um so we did have someone who said that at their job they weren't allowed to upload Claude skills. So again, that's where you might need to open it up and paste the sub skills one at a time and say, I can't install this skill, but let's work through it. And then the other thing that person tried was they found they were able to somehow install it in Claude code and then call it from other chats from there. So that could be another thing. Maybe you upload it to a project and you call it from a project. Also note I am using a Claude project, and that's where all of the transcripts are. So um that would uh that would be uh an all different ways to do it. So no matter how you do it, um you can use the skill in many different ways and in different tools and systems. If you have any questions about it, especially before you buy it, make sure you get in touch with me. You can find me at deb at deltac.com. I'm always happy to answer some questions. Um and uh oh, thank you, Deanna. We've got a hello from Deanna. Oh, check it out. Here we go. Thank you, Deanna, and hello to your dog. I don't I don't use Instagram anymore. Hope dog is well. Uh, so I'm gonna wrap up for today and get ready for my other stream. If you would like to watch my other stream, please join, uh, please follow Vocal Coach Debbie on um uh Twitch or YouTube. I'm going to be going live in about a half hour to talk about uh breathing techniques for better singing. And in in reality, I use these breathing techniques in my speaking, public speaking as well. So uh hop on over to Vocal Coach Debbie, follow me there, and I will see you uh live from my vocal booth uh at the top of the hour. So thanks everybody and have a great rest of your day.