Hello. Hello. Hello. Hopefully I'm live. What's up everybody? I see we got some people here. Thanks for being here. Thanks for being on time. I'm Joe. I'm going to be leading our workshop on a very cool new release from Lyssna. One of, one of the best, one of the coolest and most useful which is the Lyssna MCP.

So thank you guys for all being here.

All right, I want you to just this is going to be our quick icebreaker here. And it's partially out of my, out of curiosity that I'm asking this, but also might come up a little bit later in our workshop as we talk through some of these skills and some of the stuff that we're obviously going to be using AI for here today.

I want you to give me just an off the top, think about it for 10, 20 seconds, and I want you to drop it in the chat. Just give me one thing that AI is helpful for in your research, product design process. I know we... I'm sure we have some researchers here, product designers, UX designers. Not to get overly specific, but if it's something towards research maybe lean towards that a little bit more.

But just give me your first off the top take on something you're personally using or finding AI helpful for in research/product design land. Presentation drafting and creation, quick iteration summaries, themes, past search results. Nice. Ideating research approaches, UX writing, options to curate.

Sweet. Yeah, pattern analysis. So we got some for synthesis. We got some for let's see, Jenny, presenting, drafting, creating. We're going to get into some good tools I built for that. Insight repositories, copy and design consistency. Nice. Awesome. All right. Now let's do one more. I want you to give me in the chat one thing that AI is not so helpful for in your research or product design process.

I left this open-ended with the not so helpful thing, but I don't know. You could think of maybe you use AI for something in your research and you're like "I'm going to go look at that on my own." Or "That's not quite right. That's not exactly what I was looking for." Maybe you do it a couple times, re-prompt it, and then you're like, "You know what?

I'm just going to go do this on my own." Anything like that, anything where you find it maybe falls g- total- just in general falls a little bit short of your needs or maybe still has some room to grow. Understanding context, sentiment, hallucinations, almost too easy to not review everything, so things get missed.

Yeah. And that's my bad for being lazy. Yeah. You know what, Melanie? If you're putting this in the chat, if you're aware that it's happening, it's, and you know that it's a possibility and not just running with it I wouldn't call yourself lazy. We're all using these tools, but the key is, when that stuff pops up, being able to go back and say "this doesn't look quite right."

Or with the hallucination, sometimes you get stuff that's just so obviously wrong or needs some checking. Only as good as the prompt I give it. Yeah. Us humans still, we still rule. We still got to, we still got to tell it what to do, at least for now. Making sense of qualitative analysis, finding what matters.

Will contradict itself unless I point out the contradictions. To define research questions. Nice. That is, Asma, that's one I- we might talk about a little bit later in this workshop, 'cause I got some, actually just some stuff I would to maybe pick your brains on after we get through everything today.

Interpreting data, hallucinating, fine-tuning details. Fine-tuning details, that is a big one for me, Julian. I think that's a lot of stuff with AI, but definitely with research when you're like, if you're using it to get kind of the bones or the structure of a study together or something like that, it can obviously be really helpful there.

Or if you're synthesizing, it can be really helpful in getting like the lay of the land. But then when you get into the fine-tuning, the context, all of the other the stuff that really matters to get right with research that's where I see it fall short or start to get frustrating for me, where I'm like, "Why am I not just," or that's usually when I do say, "Let me just let's put Claude away for a little bit." The more specific you want to be, the worse it is. Totally agree. Cool. Thanks for sharing that stuff, guys. We're going to come back and hit on a couple of those things as we get into our workshop and, using AI for research.

So let's do it. I'm Joe. I'm going to be your workshop guide today. I'm the Design advocate here at Lyssna. I've been a product designer and a researcher for well over 10 years, maybe got to update that slide, and teaching it for just about as long. I'm going to be running our workshop today. Like I said, I set it to 12:45 so I'll try to wrap at that point, but I'll be able to hang out a little bit after if you guys have questions or, need some pointers on getting things set up.

So let's get into getting started with the Lyssna MCP. We're going to cover top to bottom, the MCP and some of the things that go along with it, including our skills library, something that I focused on building and working on, and making sure you guys are leaving here ready to use it and feeling confident about using this amazing new tool in your research and product design process.

So we're going to cover some basics first. We're going to cover getting set up with the Lyssna MCP in your agent. We're going to cover some use cases and some examples that I'll walk through with a study that I have. And we're going to install and use some key, key skills. One disclaimer, if you have a question and I don't have the answer or I'm not 100% sure about the answer I will first be upfront with you and let you know.

I won't give you a, won't give you a BS answer. I'll give you my best answer if I have it. But I will get you the answer. When it comes to the MCP, there are, lot of different use cases for different agents. I'm looking, Julian, at yours in here. And there's perfect example, Julian, right on time to get you an answer to those.

I want to make sure that I get you the right information on this stuff. There's a lot of variables, a lot of different use cases, different policies, plans just stuff that can factor into some of the answers, especially like the more technical stuff with your questions. So I will answer them to the best of my ability.

We'll cover all the big stuff. But if I don't have that specific answer, I will make sure that I get it to you. Cool. So let's get into it. Let's start with the really simple high level overview. Before getting into the Lyssna MCP, just MCPs in general. I think y- I used to teach UX design classes at General Assembly here in New York, and I remember when students would do their presentations at the end, there wasn't a whole lot of focus on like technical requirements.

It was more like a design prototyping exercise. But I remember questions we're talking like 10 years ago now eight, 10 years ago. There would be questions From other students like, how would you get that information? How would you know when to reschedule someone's appointment? Like just technical implementation questions.

And I remember the answer was always, "Oh, we would just use an API," or it'd be like the algorithm or the API. And it was just like a, maybe not a bad answer, but it was always like the go-to answer of how do we do something that technically that we don't really know how to do yet, but we could probably, figure out down the road.

It was always an API. Like MCP is like the new version of that, right? Of like, the magic step that connects and solves all these things between your tools. But in reality, MCP, Model Context Protocol, it connects AI agents to outside tools and data. It can connect Claude to your Gmail and write emails and respond and look through your inbox.

You can write up Notion docs without needing to write a whole lot. And among all the MCPs out there, you can also use Lyssna to access, query, and really get the most out of your research. So it lets your agent read from and act on, depending on different permissions, the tools that you already use, not just chat.

So in the early days, pre-MCP, you might have drafted up an email in ChatGPT, copy pasted it, tweaked it a little bit in MCP, and that connection allows you to actually, send it through there or access your email. So a lot less a lot less copy and pasting for the most part with a lot of the MCPs out there.

And it's one standard that works across many agents. That's kind of the general overview, this layer of that connects your apps and the different products you're using, as I'm sure many of you are, to your agents. The Lyssna MCP specifically connects your agent directly to your Lyssna account.

So if you've got a Lyssna account, if you're doing research in Lyssna, if you're running studies, you're going to be able to access all of that stuff directly from your AI tool your LLM. We'll get into some of the more specific use cases and features but at a high level, you can ask questions about your studies and results in plain language and get stuff back and be a- and with some of the tools that I'll show in a little bit, be able to put those into action a lot quicker and remove some of the, let's say the grunt work steps of going from, creating and doing the research to actually putting it into action. This is the best way that I can describe the before and after at least where the Lyssna MCP is now.

So for those of you who aren't using or haven't used Lyssna, this is what Lyssna looks like on the back end of study. And to get into it, you got to log in. You'll see I put my password and stuff in a couple times incorrectly here. You got to find the study that you're looking for. If you're in a big team account there might be a lot of lot of stuff to dig through there.

You got to get into your results. And then finally, if you're synthesizing something, let's say you get to dig through all of that stuff and make sense of it. Now, I will still say the Lyssna platform makes that about as easy as it can possibly be, but the MCP kind of takes it a step further.

Instead of doing all of that stuff, can say, "Hey, can you pull up that pricing study I did for Chair Time?" That's the example we're going to be using today, "and give me a quick summary of the results." And one thing I want to mention here is that, in terms of the capabilities of the Lyssna MCP, that is all reading from your actual Lyssna results, and all your Lyssna results are from real people, whether you're recruiting through the panel, whether you're using a self-recruitment link.

So it makes all of this accessible in your LLM to, ask a quick question like this about a pricing study, but the content that it's reading, the feedback that it's synthesizing, the studies that it's showing to you, those are all coming directly from studies that you've run in your Lyssna account, and those studies and the feedback that you're-- the data that you're getting from them are all from real people, again, whether it's the panel or you're self-recruiting here.

So really just being able to tap into that a little bit easier and more quickly. Even though I sped this video up, it took Let's see- 50 seconds here to do that one. So what you can do with the Lyssna MCP, you can quickly find studies and synthesize the results. You can compare results across studies.

This is one I was doing actually for the Lyssna homepage recently being able to compare some older studies of a past Lyssna homepage design versus a new one. And you can use it with other tools and build it into your workflow to do things like present your findings, improve the quality of your research, turn insights, turn a study that you ran into action items and ultimately good product decisions.

And I'm going to go into some more details with those as we get into some of the skills and some more kind of extended functionality. But basically you have access and flexibility to all of the research in your Lyssna account right through your LLM. What's coming next with the Lyssna MCP?

The short answer to this is a lot, and soon. I had to blur out our channel here because I didn't want to put all our internal messages. But I wanted to give you the idea of that there's a lot in this channel. There's a lot of progress. It's something we're really focusing on. It's something we're getting really positive feedback on, something that's really powerful for a lot of Lyssna users and something we're working on our product team, developers, me everyone involved on building out even further.

In a month, this workshop might look a little bit different in terms of some of the demos and stuff you can do. The main thing that we're working on is at the moment is write access. So right now with the Lyssna MCP, it's read only, so you can't actually create a new study through Claude.

You can view all of your studies, results, comparisons, anything that's already been, you know, created and launched in your Lyssna account. But at the moment publicly at least you can't write or create a new study through it. That is something that is coming soon. I won't give any, I don't have any dates or anything like that, but super, super high priority on this.

So I'll be sure to keep up with you guys after this on updates there. But for today, we're going to focus on what it does right now, which is read capabilities. It can look through your studies and work, craft them to whatever you want it to do. All right. So let's let's get set up with this.

So I'm going to share a link in the chat first that you guys should click on and maybe bookmark too. And it's going to be our complete setup guide. Let me know if you guys can open this and if you're seeing what I'm seeing here. Just give me like a, a thumbs up on my message.

What I was going to say, too, as long as you got this page open you don't need to...

If you want to follow along with me and do this together you can, but these instructions are really detailed. They're really specific to each of the different agents that you might be using. If you want to I'll give us a little time to get set up on this and walk through it together.

If you want to follow along and at least, get started with it today, go for it. If not, if you just want to watch and see some of the examples and stuff like that I'll show after, and then take your time to get it set up after this workshop, you'll have all the steps written out for you there.

If it's blocked or something like that I will get you an answer in the follow-up email, and you can just follow along for right now. Okay. So let me go over, And thank you Julian. If any of you guys have any of these issues, put them in the chat, and I'll make sure, I want to make sure I get back to all you guys on them.

Okay. Outside of some of those blockers here's the basics of what you need to use the Lyssna MCP and skills. First one is a paid Lyssna account. I have included in this MCP workshop setup a promo code to give you one month of growth, which is our paid plan, for free. So that means no matter what plan you guys are on right now, if you don't even have a Lyssna account right now you're going to get a month free of that plan, and I included instructions down here for each of those scenarios.

So if you don't have a Lyssna account, you can just go through. I got a... Oh, my screenshots didn't load in here. You can still follow along with it. And if you have a free Lyssna account, to upgrade with the plan here- And let me save that, Sammy. And all the instructions are here. So I tried to go from the very top with these instructions of what kind of account you have and then we'll get into connecting Lyssna to your AI tool.

So you need a paid Lyssna account. You need one of the supported AI agents right now. We have Claude, ChatGPT, Copilot, Gemini, and Cursor. I'm going to use some... I'm going to use Claude for some of the examples here today. There are also some slight differences in how to install things on if you have a free Claude plan versus an organization one.

That stuff is all detailed in the guide as well. But I guess for starters, you need one of these agents. This is what's going to work and connect to the MCP. And you need studies in your Lyssna account to query. So if you're new to Lyssna, and you don't have any studies or you're just getting set up right now you can just follow along.

I have some examples that I'm going to use, but the, you can connect the MCP without doing, without having any studies, but it's not going to be super useful just yet if you don't have any, you know, results or responses or studies to to query or ask questions about. So these are what you need to get started.

Those are all in that guide. There's also some stuff about skills down at the bottom, which I will share in just a little bit, but all the instructions are there, and you can just start... I know it's a little dense, but you can start from the top, and I tried to make it as complete as possible for all the different possible scenarios.

I will also, for those of you guys, if it's blocked by your IT not only will I get you guys answers on that or figure out a, solution there I'm going to update this guide and the workshop link so that we'll so hopefully we'll be able to, avoid those issues or at least give a, a workaround or instructions in future ones.

Okay. So let me I'm going to give just a couple minutes here to for you guys to get connected for those of you who want to, want to get it set up right now. I'm just going to put three minutes on the clock here and let you guys rock. If it- if you're blocked or if there's anything else standing in the way, let me know in the chat.

Otherwise I'll give you a little time to follow along with those instructions and get started. Open the instructions on your phone. Nice.

So yeah start up here with what fits you. Or the one that I guess best describes you in terms of a Lyssna account. First step is getting on on the paid plan or a free month of the paid plan. You can use this promo code. Now, my manager sent me this promo code with this exact caps, like MCP, capital, F-R-E-E, lowercase.

So I put it in here exactly as it was sent to me. I should probably check if it if it works in any... If it's case sensitive, but I would just go with what I have here and let me know if it's working or not. And then we have instructions down here for Claude, ChatGPT, Cursor, Gemini depending on your plan and and some links there to get it connected

Finally, if you are connected, let me know in the chat. And we have step three here to ask your first question just to make sure it's working and pulling some of those responses or some of those studies from your account. So let me give a couple minutes for that. If you guys have any questions, let me know in the chat.

If you do get it connected and, get some actual results for a question like this let me know in the chat. And if for whatever reason you're not getting it set up now or have some, some kinks we need to work out, then I need to get back to you on. Just chill for a little bit and we'll go through some examples in the skills right afterwards.

All right, my timer just went off, but let me give you guys another minute or so.

Let's move on here. And as usual, guys, let me know in the chat how you're doing on this or if you have any questions or things popping up. But I'm going to go through it on my own with a little test drive here. So I'm going to walk through this with the MCP connection, which I already have set up.

And then we'll get into some of the skills before we wrap up. So just to give you a little context around the example that I'm going to be using here I made up this fake website, web app called Chair Time. I made a little mock-up of it. And for these examples and demos, I'm going to pretend that, I don't know, I or we are the research/product design team for Chair Time.

So Chair Time is a booking, scheduling, and payments app for independent stylists, barbers, and beauticians. Clients book themselves, reminders send themselves, and deposits land before they sit down. There's got to be plenty of stuff out there like this. The one I keep thinking of is Mindbody, which is what every fitness studio and stuff uses and it's really bad.

So this is the one I'm using as our example that we're doing some research on. So that's what Chair Time is, and the example study that I'm going to be using to demo some of these skills is a pricing study. So as a team, the, let's say, Chair Time research product design team, we have this standard plan that's 20 bucks a month.

That's what stylists, beauticians, barbers are paying for, all the features and scheduling and stuff in the Chair Time app. And let's imagine that we have a a new feature idea. Maybe we're getting a lot of feedback from customers our stylists, barbers, that no-shows are a big problem, right?

People sign up, they book an appointment, and then they don't show up, and that, customer loses on a slot they could have had filled. They either have to, as I saw in some of the research, be the bad guy to charge them or, keep their card on file. So let's say we've identified this as a pain point, and we're thinking about adding this feature called no-show protection, where you can set a fee- and automatically charge it if a client is a no-show.

And what, as we'll look at in this study we're trying to test out, is the pricing of this add-on. So maybe we know this is a pain point, we know it's something our users want, or it's something that is a challenge for them something desirable for them. But we want to see, as an add-on to this 20 bucks a month what people would pay for it.

So I have a study set up here that has 50 responses for this. And I'm not going to go through the whole study here, but it's a little screener, a welcome screen a task for them to look at that pricing screen, and then some questions the, what is it? Van Westendorp price survey to ask them, what would be way too expensive?

What would be a bargain, a great buy for your money? What monthly price would start to feel expensive, but you'd start to consider it? So pretty much a, a pretty, relatively simple focused study meant to put this new feature in front of our users and get some feedback on how we might be able to price it.

So this is all in the Lyssna account, but I'm going to use this as our example for some of the for testing out the MCP here. All right. So let's let's do it. So there's my chat right here. Can you pull up that pricing study I did for Chair Time and give me a quick summary of the results? This is a good first prompt to make sure that the MCP is connected, that it's pulling your studies in, and that it's getting the right information from your Lyssna account.

So I already put this one in but you can see that, did some thinking and connecting to the Lyssna MCP and we've got a pretty quick, but still pretty detailed summary of the 50 results from this study here. Now, I'm not going to go through all of this and I'm going to show you with the synthesize research skill how we can even go, more detailed and more reliable in terms of synthesis here.

But just to give you an idea of the workflow, we've got the study created with 50 responses. We've got a specific goal for the study, which is figuring out the pricing for this new feature. And in terms of the MCP, we've got it connected and pulling our actual studies and results right in here with some context and some I guess next steps around it.

So you don't have to use this exact prompt, but a good one to start to just make sure it's connected is say "Find a study and give me a quick summary of of the results there." More so a good prompt just to test if it's working than to actually get into synthesizing everything. Okay, let me pause there for a second and see If we have any questions or anything else in the chat.

Okay, so having promo code, that's from the chat. Let me pause there, and then I'll get into our skills. I'm going to take 10 minutes on the skills, so we'll go a little bit over here. But let me know in the chat if you were able to connect and query your research or if you have any questions on how to do that.

Okay, cool. So that is yesterday only, don't have any sessions, MCP tools to create a research task.

There is not just yet Harshal, that is coming. Write access is coming. In the meantime it's read access that'll be able to find your studies and results. I could give you a public link, I think, to a study. It won't work exactly the same way, so I'm hesitant to do that. Short answer, no, there are not tools to create a research task directly into Lyssna right now.

It's coming very soon. All the stuff we're talking about here is reading studies and results that you already have. Okay, cool. Let me let me move on here. That is the Lyssna MCP, which basically creates that connection between your LLM and your Lyssna account. But one thing that I wanted to share with you guys beyond that are

skills.

Skills in general, as I'm sure a lot of you guys are using them in different capacities, are basically just like reusable instructions to teach your agent how to do not just a research task, but any task well. I have skills set up to put format what I did during the day into my team meeting notes, stuff like that.

Guidelines and specifics and rules and workflows to not only, just trust your LLM to do it raw, but to guide it with some further instructions. In terms of the Lyssna MCP, they pair with them to turn some of the raw study data into useful outputs. And we have five free skills that I basically created.

Ask your research anything, a study audit, synthesize research, turn findings into action items, and share your findings. I'm going to highlight a couple of these right now and just show you how they work. But the skills are all-- it's on the same workshop page, but I'll send you The detailed version.

Thanks for the details, by the way, in the chat. That covers all of these skills. So basically, with these skills my goal in creating them was to use expert research, my research brain to improve how AI helps you do these tasks that are really common, really useful, and really important to get right.

So the idea here with all of these skills is to infuse some of that ex- some of my expertise and experience and best practices into the prompts so that you're not just feeding, general Claude a research study and asking to synthesize it but giving it some guidelines to do that well, to provide context to you, and to look at that research as a professional researcher would.

Really, my main goal is, in creating these, was to not only make it easy, but to make it reliable, sound research and not an AI hallucination. So let me show you a couple of these skills before we wrap up, and I'll demo a few of them and give you some ideas on how you can use them on your own.

Like I said, the instructions to install the skills are in that link there. Again, if you just follow everything one by one, there's instructions for each of the agents and, different scenarios you might need. Okay, so first skill out of the five that I wanted to highlight is synthesize research.

I would say this is the top of my list when I was thinking what would be some useful skills for researchers who have a bunch of studies and a bunch of responses in their account. So synthesize research turns raw study results into sourced findings, insights complete with evidence and the minority view beyond, behind every one.

So let me show you how this works. I'm going to do synthesize research. Synthesize this Chair Time study. This will probably take a second. Here we go

Okay, cool. So while this is going I'll give you some of the some of the basics of the synthesize skill. And you can actually just read into the synthesize skill and see exactly what it does here and what it's some of the guidelines it's giving to Claude or your LLM when asking to synthesize a study.

But I can give you the high level ones since I created it. Main thing is it is not making any assessments or it's any, let's say, judgment on the study results. I wanted to keep this synthesize skill something that really just looks for tensions, themes and gives kind of a raw synthesis of it without any interpretation, right?

So if you are using this study to then think about some next steps for prototypes you want to test, that would involve some further work after this. Synthesize is really... I really wanted to keep it focused on doing a good job of synthesizing the results, not missing anything, definitely not hallucinating any results, and also providing context or sources, let's say, as to some of the things that it's surfacing for it.

I can share this chat with you guys, but it's gone through all 50 responses. This is a Van Westor pricing survey for Chair Times no-show protection. We get some headlines. People like the idea but won't pay much for it. The tensions, deposits versus client friction, no-show frequency.

The study's still marked as recruiting. It's a self-recruited link. Almost every answer is logged in 1.2 to 1 point se- q- seconds per question. That's 'cause I used some dummy responses for this one. So the idea here is this is what a readout is going to look like when you use the synthesize research skill.

And it's going to be different for any of the methods that you use or any of the studies, but I wanted to share it as a good example and one that really tries to focus on accurately pulling all the different variables that might be important in synthesizing research without doing what r- regular Claude does a lot, which is finds a couple themes to latch onto, really makes those seem like, front and center, and buries everything else.

So a reliable skill for an in-depth synthesis of any of your studies. Let's go to the next one here. Share findings. Okay, this is my favorite skill that I created, and the one that's saved the most time for me, and hopefully will save the most time for you guys. As researchers, and I'm sure some of you have been to this workshop that I've done on presenting research findings to your team and stakeholders one of the more challenging things, at least I find, is doing all this great research and then having 20 minutes to share it with your team, stakeholders, get buy-in, and figure out next steps on it.

All of that stuff that went into it, all of that feedback, when you're presenting it or, preparing a presentation for it, it gets tricky. You have to leave stuff out. You have to focus on the big stuff. You have to focus on the things that your team and stakeholders are going to care about. And among all of those challenges, they all lead back to one thing, which is being able to condense a lot of research and a lot of complex or nuanced feedback into a, in, into communicating it in a way that is going to work for whatever your purpose is, whether that's a quick slideshow that you're sharing with your stakeholders, whether that's a FigJam that you are creating to discuss with your team.

So this skill, share findings, is all about packaging up synthesized findings in a way that for whoever needs to hear them or see them, in a way that's going to work for you to then present them to your team. So let me show let me show what that looks like for this one. So we have this Chair Time feedback here, and let's say that we want to share findings, right?

Turn, I want to create a, let's say, Chair Time branded short deck to present this study findings to my stakeholders.

So with this skill it'll ask for some more context, right? So tell me what's behind this one. Who's Chair Time? Who's the audience? Let's say Chair Time is a app for stylists, ah, here we go, barbers, independent beauticians. It's hour. Let's say in this example, I work there, and let's say we have identified no-shows as a pain point for customers.

Sorry, I'm typing so bad here. Pain point for customers, developed a feature concept, no-show protection, and want to test out the pricing as an add-on. So this is an important part, and one that I wanted to build into the the Share your findings which is providing some more context about the research and what you're trying to present with syn...

I meant to say, "Looks good." Wow. With the synthesize research, that one is all just all about, no context, no goals, just understanding what the research is telling you, the themes, tensions that are popping up. But as you guys know, when you're presenting research to a, stakeholder team, et cetera there's...

It needs a little bit more to be able to know what to focus on. So right now it's building that out. Let's see how it looks We'll let that rock for a second and see how our slideshow comes out. Now, the idea with this skill, you don't just ha- it's not just for building a deck. You could ask it to write a quick Slack summary to your engineering team.

You could ask it to create a FigJam for to discuss next steps with your product team. So it's very flexible in terms of what it can do, but the overall summary of this skill and, why I find it to be one of the most useful, is that it's packaging up raw research synthesis into a, I'll say loosely, presentation that you're then able to share with your team, communicating things the right way, focusing on the right key points, and saving you some time and, packaging it up into a nice into a nice setup.

Okay, so this didn't use the Chair Time branding as well, but I also have my Lyssna branding stuff connected too, so that probably doesn't help. But just looking at the content here, and even in the chat, you guys can see this, ... It did a good job of cutting this down, right? Not just taking and dumping a big raw synthesis and study results and every bit of feedback that happened in here but being able to tailor that a little bit with some context and cutting it down into something that's going to be more palatable, more actionable for your stakeholders.

All of the structure of this study were was created by me and how I present things and how I some of the more effective ways that I've found to take a big research project and a lot of results and share them in a way that's both accurate, complete, and trimmed down enough to be actionable.

So I forget who said it at the beginning, but let's see, saying using AI for presenting. This is let's see. Where was that presenting? I think it was Jenny. This is the skill for you here. Okay. Next, I'm going to skip the spotlight on the next one just for time in case you guys have any questions since we're going a little bit long.

The other one that I was going to share that you guys can try out is the Study audit. So this is a skill that will check your study before you launch it. The reason I created this one is because I work with a lot of Lyssna customers that I help, build their studies for and launch them. And I don't know if you guys experience this, but there's always like kind of a, what's the word?

Analysis paralysis before launching a study, right? You get the buy-in to do it, you get the okay, you build it, you tweak it, you change the questions a million times. And I find that with myself and with a lot of customers that I work with you just want a second set of eyes on your study before you blast it out to 100 people and make a product decision on it.

So this skill, the Study audit, is helps you do exactly that. It helps you catch m- little mistakes in your study or things that could be reworded. It helps you catch questions that could be biased, gaps, unclear questions and question choices. Basically acts as a second set of eyes to offer kind of a, a analysis and a suggestion for ways you might improve your study before launching it.

Unlike most of the other skills that are involved, that mostly involve synthesizing and packaging up or doing something with results, the study audit is a skill you can use before launching a study. Just like a, the other prompts in Claude, you could say, "Hey, I just created this Chair Time study.

Can you run study audit to take a look before I launch it?" And gives you some options to consider, things to ignore. Maybe there's some tips that you don't want to include but mostly gives you confidence that you got everything covered and aren't launching something that you're then going to have to, redo and launch again.

So those are three of the five skills. All of those skills are in the startup guide and there's instructions to install them all at once or one at a time on each of your agents. The best way to get familiar with them and see how powerful they can be is to use them on your own. But wanted to give you a little taste of some of the some of the big ones and some of the ones that I use really frequently.

All right. What to do next? Get the Lyssna MCP and skills installed. Try a few skills on your own studies and modify the skills to fit how you work and, your own processes and internal guidelines, stuff like that. A couple ideas for that is in terms of modifying these skills is you could add your brand voice guidelines to the study audit skill.

So if your company has tone and voice guidelines or, UX writing guidelines and you want to audit a study, you can add something into that skill to not only audit for the things I just mentioned, like bias questions, but also to make sure that the tone, voice vocabulary that you're using is in line with your company's guidelines.

For the share findings skill, if you have a weekly meeting with your team where you're doing stuff like this, maybe it's a FigJam, maybe it's something you're putting in Notion or Slack you can provide that and have the skill format the output into the way that you're actually presenting it rather than just a, a raw slideshow like the one I shared.

Connect it to your analytics tool and synthesize and use the synthesize research skill to even add more context around a study or to compare a synthesis from a Lyssna study with some of your other analytics or metrics there. These are just a couple ways to get you guys thinking about this but wanted to share them to say that, while the MCP and the skills will get you really far in terms of time saving and efficiency and proficiency with with your research, that there are also ways to take them even further and make them work for, your day-to-day research tasks. Cool.

Okay, let me pause. I know we're way over on time. I'm going to put this link in the chat for feedback here in case you guys got to run. But let me check the chat for some other any other questions we have here. So Theresa said "Does the audit skill read the copy that are baked into the image, images that I upload to a preference test?"

That is a good question. I believe it can. The... One of the things that I had... if not, you can upload those into... You can add those in. I'll check and get a, a 100% answer for you. I know one of the things... The only thing that's giving me pause, Theresa, is I know that one of the things I was working on when creating these skills were to look at the images for a preference test and make sure that they were equal fidelity and good resolution, which can throw off a study, and it had some issues with that.

So in terms of actually reading the copy in an image, I am not sure. I know that if you just drop those in, additionally to this study audit skill then that would work just fine. But right off the right off the top, I'm not 100% sure. I could check it, though, on this. Let's see.

"As I'm from Belgium, we're doing multi-language studies. We make duplicate of the test and translate it. It might be an edge case, but is this on the skills radar to combine multiple studies into one report?" That is a really good good question. It is... So we have... to answer the question directly, it's the skills radar we're waiting for the right access.

That's going to be the next big release for the MCP and skills. And before that's like inside, inside Lyssna. But before that happens, we we have a big menu of possible skills and things that we want to do that we're going to act on really quickly once that's up and running.

So it's not, it is an edge case, but you would not be the first person who's asked for something like this. So this could be, you could combine the studies into one report with the skills that you have right now. If you were to say synthesize and give the context that one is in one language at, or multiple studies, you can do that right now.

But it is something it's not on the it wasn't on my radar before your comment, but it's something I will look into and something that's probably, you're probably able to do or modify with the skills that you have right now. Let me save that. That's a really good edge case too. I'm going to look into that on my own and also check with our product team.

Let's see some other questions here. One question, will these skills or future ones work with interviews recordings, transcripts? Yeah, it will get the transcripts from them. I did a really simple kind of like survey setup for this one. One, 'cause it's a little bit quicker. But yes, it can get interview transcripts and, pull quotes, themes, et cetera from that.

And does, does pretty well from recordings. Let's see. Will the recording be available after? Yes, it will. I will be sending it out. I'm going to put together some responses to some of these questions and I will send out an email with the recording in like the next couple hours. So yes, keep an eye out for that.

I think also if you just go back to the link you used to, Access, like the live workshop here, you'll get to the recording. But either way, I will send out an email with the recording link and all the resources from this. See some other questions. Hannah, we're also doing multiple copies of the same study for different segments, for sample different markets or age groups, not an edge case for us.

Yeah. No, it's... That's what I was thinking. Thank you for sharing that, Hannah. The it's not an edge case. Maybe like language translation and combining, showing those in a single report. The language aspect might be an edge case, but it is something I have heard and experienced plenty of times, which is like slight variations on studies that you want to be able to-- that are marked as separate in your Lyssna account, but that you want to, report on basically as either the exact same study translated or a the same study for a different market.

So what I would say is it-- I would love your feedback on this, guys. You can try, you can synthesize multiple studies, whether they're different language or different segments. You can do that with what I've showed you here with the MCP and the synthesize skill. Might have to give it a little bit of context to just share like what your end goal is.

But you can do that with this. You guys should give that a try and let me know let me know some feedback on the responses since we're going to be not only creating new skills, but updating them quite a bit based on feedback. Yeah, Julian, respect, experiment. That's the way to do it.

Audit skill, big tense. Cool. Any other questions, guys? Thanks for hanging with me a little later than usual. That's always the case though, but I got to get to all your questions. I think we're all set here. Let me drop this link in one more time. If you guys could drop me some feedback on the workshop, that would be greatly appreciated.

You can answer like literally one question just to see how things went today. Other than that, thank you guys for being here today. Appreciate all the good questions. Appreciate your interest and willingness to experiment with the MCP, and I hope it was helpful for you. I'm going to share everything in an email, recordings, guides, all of that stuff.

All the questions in here that I didn't get to or like some of the specific questions, I will either get to you individually with a response or for some of the maybe some of the other ones that are more widely applicable, I'll put some answers in the follow-up email and notion as well after I look into them.

But either way, I will make sure I get back to you guys with with some solutions to some of those. Let's see, hang on. Feature I'd love to have, automatic quotas for participant recruitments. Boom. Saving that. I love these suggestions. But as you guys can probably imagine, just working at Lyssna, a research company, we take feedback seriously.

We build a lot of stuff that happen in like workshop chats here. So all of all these suggestions and just feedback here, like I'm marking it. I'm going to look into it. I will get it to to our team. And it's good timing too, since we're, building all this, this new MCP and skills stuff from scratch too.

But I appreciate you sharing it and know for sure we build a lot of features that just come from conversations and workshop chats like this. So the more we know, the better, better we can do to hopefully send you an email soon and say, "It's done. You can do it." Thank you guys so much. Have a great rest of your day.

Thanks for hanging with me. Have fun with MCP. Get that free month and crack into it, and I will follow up with you guys shortly with everything from today's workshop. Have a good one, everybody. See ya.