Ramli John

How AI is changing user onboarding, one level at a time (with Ramli John of Delight Path)

How is AI changing user onboarding, and what should SaaS teams actually build? In this episode of the Love at First Try podcast, Jim Zarkadas sits down with Ramli John, founder of Delight Path, creator of Product Leaders Lab and author of two books on onboarding, Product-Led Onboarding and EUREKA. Ramli helps product teams turn more of their users into happy customers, and he runs a peer community for product leaders. Below are the key insights from the conversation, followed by the full transcript.

The four levels of AI in onboarding: tells, guides, generates, acts

Ramli sees AI being used for product adoption at four levels, each harder to implement but better at getting users activated. Level one, AI tells: a chatbot that ingests your knowledge base and answers “how do I do this?” with instructions. Level two, AI guides: when a user asks how to do something, the AI launches the right in-app tour or hotspots, which he sees tools like Appcues, Userflow and Pendo connecting to user requests. Level three, AI generates: in open-ended products like Notion or Miro, users describe what they need and the AI builds a custom template instead of making them pick from a list. Level four, AI acts: the agent does the work on the user’s behalf after they approve it. Ramli’s examples are Salesforce’s Agentforce updating and reassigning records, and QuickBooks testing invoice creation.

Acting on the user’s behalf cuts the time to the aha moment, with a trade-off

The fourth level is still at the edge, and Ramli isn’t sure how popular it will become. Its upside is that it cuts the time and number of clicks needed to reach the first win. The catch he points out: when the AI keeps doing the task, users may never learn the pattern inside your product and become dependent on the agent to do it again. It’s also the hardest level to implement. As Ramli puts it, you need the engineering, the API and an AI that knows your product and integrates with your code base. Some companies build it in-house, while others use third-party tools like Foldspace for generating and acting.

Delight is exceeding expectations

Ramli finds “delight” a vague term, so he defines it through the user: users arrive with a set of expectations, and delight is what exceeds them. His example is the first time people used Uber over a decade ago. They expected to wait to flag a cab and pull out a credit card to pay, and instead they just got in and got out. Taste, in his view, is learned like food: you try a lot of things, store what’s good and what’s bitter in memory, and use that experience to decide what is good, great or bad. He thinks AI can learn some of that, but at the end of the day taste is based on human experience.

AI is an average, so talking to users is your edge

Across the product leader summits and groups he hosts, Ramli says most AI discussions come back to the same point: teams need to talk to users more than ever. In his words, AI is “essentially just an average of whatever’s online”, while the insights outside that average come from real conversations. He believes you can learn more in a five-minute conversation than from a thousand surveys, and those conversations tell you what users now expect, such as products having chat, so you can figure out how to exceed it. He also sees more companies, especially AI companies, building in-person communities because human connection is something people can trust.

One audience’s delight is another audience’s annoyance

Building on Jim’s example of ZenMaid, which serves cleaning businesses, Ramli explained that delight is not universal. Notion users may love a little animation and guidance, while busy cleaning business owners may find confetti annoying, enough to actually leave the product. That’s why he keeps returning to connection: understand who you’re serving before deciding what delight looks like for them.

How to train your taste

Ramli’s advice for founders, designers and developers starts with experience: look at the market and analyze what you like, what you don’t, and specifically why. Second, stay connected to your customers and their emotional responses. Third, connect with peers. He compares it to chefs and foodies who learn the nuances of a dish from each other. With the hype and pressure around AI, he believes connecting with other product people matters more than ever.

Just pick one: an activation story

Ramli shared a story from a marketing and SEO tool he worked with, without naming the company. Its product could do many things, and new users landed on a dashboard asking them to do around ten things at once, so they froze, got overwhelmed and left. Ramli talked to users to understand why they came in the first place, for example to write an SEO article or to crawl their website for issues, and identified two or three core jobs to be done. New users were then directed to the specific flow for the job they wanted. He doesn’t recall the exact activation numbers, but the impact was significant enough that the team was happy with the result. His takeaway: it’s hard for founders to pick one, but focusing what you show is what works.

Find jobs to be done in 5-10 onboarding calls

When it comes to finding those core jobs, Ramli finds it hard to do quantitatively by looking at pages and events. Instead he digs into customer onboarding calls and asks why people signed up and why they bought, trying to recall the first moment they started looking for a solution. In his experience, five to ten calls is usually enough to identify where users want to start.

Ramli’s AI workflows: Granola, Claude and ClickUp

Ramli records his calls with Granola and used to copy and paste them one at a time, until he connected Granola to Claude through an MCP. Now a daily agent prepares him for upcoming meetings with context and next steps. A second agent pulls from his peer group sessions and suggests, on a weekly basis, LinkedIn posts and articles to write. He’s also connected Claude to ClickUp, so client, customer and community member records stay up to date after each call. His favorite software is Granola, which he even uses on mobile for everyday meetings.

Full transcript

Introductions: who is Ramli John

Jim (00:26): Welcome to the podcast officially and thanks for making the time today as well. As I mentioned, we always start with a quick intro of who you are, what’s your story and what are the people and the teams that you are currently helping and how as well.

Ramli (00:40): Yeah. My name is Ramli John. I’m the founder of Delight Path. I’m also the creator of Product Leaders Lab. And so I work mainly with product leaders, as the name suggests, and product teams. on one end I help product teams turn more of their users into happy customers. I’ve written a couple of books, Product-Led Onboarding and EUREKA. And then on the other end I’m working to build a space and community, a peer community for product leaders, ’cause as we know, AI has changed the world. specifically it’s changed the world for product teams where, you know, AI can do a lot of things now that can accelerate things. So that’s that’s a little bit about me.

Jim (01:22): Nice, yeah, and fully agree with AI. I can really see like communities have been a big trend for a while on the internet, but nowadays, especially with AI, like the positioning that you have in your community, if I have to be honest, because the… Yeah, it’s not just about bringing people together. It’s also about solving a specific problem that they’re all experiencing. It’s especially like as a head of product. It’s not only how you’re gonna put like integrated AI into the team, but also what should you build with AI. It’s so hard to separate the hype with what actually makes sense. So yeah, I really, love what you’re doing there, honestly.

How to define taste

Jim (01:54): Yeah, and the first topic that I want to dive into because your company is called Delight Path, something that I always liked, is how do you define taste and delight? And we can analyze each term separately, like taste and delight. So the first one is what do they mean and how do you define them?

Ramli (02:15): Yeah, whenever I hear the word taste and it’s become more important because it’s the thing that really you must have heard this in your show before, that it really differentiate what is good, what is great and what is bad in terms of output with AI. Output of anything. But AI more so because it’s it’s generating things and taste is really understanding and looking at it and deciding Is this good? Is this great? Is this terrible? Does this make sense? And obviously taste is something that you get by experience and the word taste itself is related to food. And how do you know what is good food and what is bad food? You try different you try a lot of things. It’s experienced. I was like, that is bitter. So when you taste I don’t know, something bitter like bitter melon. and it’s better, so it’s now stored in your memory. So it really is looking at the world, taking in all this experience and deciding for yourself, for a community, for your users, for your team, for your company, what it would be good or what would be bad. and I think that’s something that a AI can kinda learn, but at the end of the day it’s based on a human’s experience. and I guess you’re trying

Delight means exceeding expectations

Ramli (03:42): to tie it back to delight because delight is that term that so vague, right? It could be a lot of different things. It is very subjective in a sense, but it does come which is similar to taste, but it does come really back to what when I when I think about delight it comes back to the user, right? What is the thing that surprises them or gets that emotional reaction or, you know, that moments of magic or What is, you know, that thing that really exceeds the expectation that is probably more of a better term for delight is you’ve they have a set of expectation and now you’ve exceeded it beyond what they had in their mind. And at that point they’re surprised because it’s like I’m expecting this. for example, the first time people used use Uber way back then, over a decade ago, you expect it take a while for you to flag a a cab but and you have to pay and you have to pull out your credit card but the first time you use Uber it’s like I just get in, I get out, I don’t have to pay, it’s like, wow, that’s exceeded my expectation. So that’s how I would kind of look at Delight and

Jim (04:57): Yeah, we love that. That’s something that Amar, CEO of a team, of ZenMaid, one of the teams we’re designing for, he said that the way that he sees product light is an unexpected surprise, which I really like that there is the unexpected part and the surprise part as well. But what I like with your definition is also like, it doesn’t have to be about the design, right? Of course, if we think about product interface, then it would be, I don’t know, like the most mainstream examples like confetti that I don’t even, like they’re too predictable right now because they’re overused.

Ramli (05:25): Yeah. Right.

Jim (05:28): But the example of Uber is really good because sometimes about the product impact, like one idea, for example, that Amar said that he wished, like at some point he would be able to do is that people sign up, let’s say for ZenMaid and you get a lead from ZenMaid. within the first week, let’s say. So that’ll be something

Ramli (05:42): Hm. Yeah.

Jim (05:45): unexpected. I got a new customer just because of ZenMaid, let’s say. So it’s a very interesting thought, of an exercise actually to think about what would delight look outside of product, like a surprise in terms of the value that you deliver or the expectations that they have, like the level of value and the level of impact that you do have. Yeah, that’s a really, really nice one.

Why AI pushes teams to talk to users

Jim (06:10): Do you, because with the AI, I’m sure you’ve seen a lot of hype online. I mean, LinkedIn is a kind of a crazy place if I have to describe it in a way myself.

Ramli (06:15): Mm. Yeah, it’s so crazy.

Jim (06:21): Yeah, with a lot of noise and it’s a that I personally don’t like about social media, but that’s for another episode. And what I’m curious is… If you feel like the whole, they’ve been saying that taste is a new moat, like, I wouldn’t even go into that because I feel like the whole thing with the moat is like too big of a discussion and sometimes people try to define something as a moat. I would say more if you believe that. Taste is a way, taste and delight is a way to differentiate more and more in the world of AI because there was these hypothesis that more products come to the market. Many of them are just vibe design with not so much thinking. So like delight is becoming more and more important to stand out from the competition. But do you, is this something you’re experiencing?

Ramli (07:08): It is it is something I’m experiencing because at the core of delight and user delight and product delight is understanding your users. Yes, talking to them. The majority I’ve talked to dozens of product leaders and even in I’ve hosted product leader summits and I don’t I don’t kind of plan this out, I don’t say it, but a majority of the discussion around AI goes back to hey, we need to be talking to users more and more than ever. Like not like texting or looking at data but actually having this conversation with them because that first of all the human connection, like you know that’s real. It’s hard to turn Jim or somebody into a video call that you can’t you can tell it’s like something’s off. AI video is not there yet. It’s not. Like you can say something’s off with that face or that hands. Do you have six fingers?

Jim (07:55): Yeah, 100%. Yeah.

Ramli (08:01): So once again, us in this in this conversation is like we’re we’re having this human connection even though we’re hundreds, thousands of kilometers apart from each other. I think that brings it back to that. But at the second time, you can learn so much more in a five minute conversation than you can in like a thousand surveys often or looking at the data. And I think why this is all important is because having those conversation, those user research, those insights helps you set, okay, this is what they’re expecting. They’re expecting now That products will have chat, right? Or they will have this. How do we exceed that? How do we make it so that we go beyond that specific expectation that they have? And so I think that is the differentiator, first of all, the human connection. It’s like, we’re real. I’m seeing more companies, especially AI companies, build inhuman, in person, in person communities, real life, because it’s like, How do I know I’m not talking to virtual Ramli or AI Jim or somebody somebody else? Right. Crazy times, right?

Jim (09:10): Right, crazy times, yeah.

AI is the average, human insight is the edge

Ramli (09:13): Well I don’t know. If I’m texting Jim, is it is it him? Is it AI? Something else? But when we’re in a call like this or in person, it’s real and I think that’s first I see companies are differentiating and what that brings is a ton of human insights that you that AI can’t it just AI is essentially just an average of whatever’s online and maybe there’s insights there’s there are insights outside of the average on in the world and that you can truly bring into your product design, your user experience and everything else.

Jim (09:46): Yeah, yeah, I love it. And what I really like, like the way that you approach it, even took a note now, like on my mic that’s like delight is about exceeding expectations. That’s how like, that’s a phrase that keeps coming to my mind, like the way that you describe it. And I really like this because for me, What really drives me and what I really love as Jim is quality. And I’m into the product design space. So like crafting a product that just feels like sophisticated and well done. is some love in it. That’s something that really kind of inspires me. And that’s what I love with your definition, the delight part is like, and also on the, AI kind of an impact, you could say that it’s kind of a pushing the industry to exceed expectations more like to, to push the limits and to really start thinking about details and not stick to the average, which is a good thing because we’re just going to get better products. That’s one way to see it. Like other people may say that now you’re just going to be flooded with average products, but I think both are going to happen at the same time because of the many average products they will need

Ramli (10:43): Yes. Yeah.

How to train your taste

Jim (10:45): to stand out more. So you have more good ones. and quick question, because you’re a researcher, like you put a lot of time into research, uh, for producing quality content and all these things. And I’m curious, how would you advise, let’s say founder or a product maker, could be a designer or developer, how do you advise them to train and improve their taste and delight skills in a way, either any way, or like at least how do you do it?

Ramli (11:16): Yeah. I would say it goes back to the tasting the food, this experience, looking at the market, what do you what do you like? What don’t you like? I think analyzing things that what you like and don’t like is important. And why specifically why you don’t like I think one of the questions you’re gonna ask is was what is my favorite SaaS? I’ll tell you I’ll tell you that when that that comes up. The sec the

Jim (11:36): Yeah. Yeah.

Ramli (11:39): second thing is it goes back to having that connection with your customers and or your users or that’s so important. I think that’s that goes back to understanding what do they like? What what is getting into their mind, getting into what is their emotional responses to things is important. I’m gonna keep going back to connection. I think that human connection is important. So human connection with customers. I think human connection with the industry, I think it’s important now more than ever to connect with other product leaders, because they’re You mentioned already the hype on LinkedIn is real. It’s just absolutely crazy. All these influencers telling us that we’re not gonna have a job anymore because AI will co design our products and yet we see [unclear: likely OpenAI and Anthropic] hiring salespeople and designers and coders like crazy, which is the irony, because we thought that they would they would replace us. But connecting with our peers, I think, is another way to build that taste. and and you can you can gain taste from somebody else. yeah, I keep going back to taste because when I think taste, I think chefs and foodies, you can hang out with other foodies and other chefs to understand what nuances of that wine or nuances of that specific flavor of that dish they liked or don’t like. And you kind of take that into your own taste itself. I’m taking that metaphor and bringing that to the product and design world where connecting with other product people and designers is important. I know we’re all I know it’s really tough. This bit it’s busy, there’s a lot of pressure, you know, you you’re you better be AI prototyping, Claude coding now or else you know, you’re gonna be out it’s really I think this more than ever to connect with our peers and really build that that taste and flavor that you that you’re talking about.

Jim (13:32): Yeah, honestly, yeah, it makes sense on the one hand, but I really like how you keep repeating this part, like the whole thing of human connection and how this can give you ideas on how to make your product more delightful, but in general to navigate is kind of a wild reality in terms of information that we’re experiencing. Yeah, I really like it because…

One audience’s delight is another’s annoyance

Jim (13:52): On the product design side, on the experience design, indeed, like Delight, sure you can look at Notion. can look at these, you can look at products that you perceive as great and world class in terms of like taste and execution. But in the end of the day, every company and audience is unique. Like Notion, if we say that this is a delightful, really well done product, is serving a very specific audience. Whereas ZenMaid, just an example of like what I’m working on is focused on cleaning businesses. And what would be super delightful for a cleaning business owner can

Ramli (14:20): Hundred percent.

Jim (14:21): look very very different from a gift that you send them to random visit to the office to kind of a, don’t know, cowork or whatever that would be. Like an idea would be just to like the other teams are like to rent a van and do the round of not the full round, but yeah, just visit different states and kind of a visit our clients to connect with them. So personally, I’m really down and I took some notes on this actually on putting it as the main topic and ask yourself, how can I connect in more meaningful ways? Now that I’m thinking out loud, one example that has helped us actually improve and add more delight, but in general improve the product is that we have one of our, let’s say best customers, we also have a marketing relationship with them because we have a podcast with them at ZenMaid. They’re joining the design call on a weekly basis. And for me as a designer, was like, thank God, like I have a user in the weekly design call. Like what else could I ask? This is beautiful. And it’s been getting back to the connection thing that you’re saying. It’s been one of the best changes we’ve done the last two years to have this person in the room because the one is access to knowledge, but the other one is empathy. Like you can understand more how they think because I’m not running a cleaning business. I don’t have time for that to develop empathy. So I need to hang out with these people more and get in their shoes and understand what makes them feel different things and what has meaning for them and so on. yeah, I really love what you’re saying and I fully agree. Yeah.

Ramli (15:51): That makes sense. I really like that pulling in the customer. I it goes back to yeah, I wanna like double down, double click on what you mentioned, ’cause what you Notion users might find delightful, ZenMaid users might find actually the opposite. They might find it annoying. They might find it like I don’t want that confetti. I’m too busy. You know, in Notion users, like I I like that, you know, that little animation and that guidance. ZenMaid customer they’re busy. They they don’t They just give me what I want, give me what I need right now. And so I think that goes back to what you’re saying that it’s it’s not just different types of delight. One you one customer base might actually find another products customers they what they find delightful us annoying and enough to actually leave the product. So I think that goes back to what you’re saying. It’s important to connect with who you’re serving.

Jim (16:53): Yeah. And on the confetti is even like on the branding side, like with one of the teams sitting into the knowledge base industry, like what they were designing, we went too far on the creative, like visual delight in the past, because we have a mascot and now there, and we went a bit too far, in terms of creativity and yeah, it started feeling too childish in a way. And like they’re selling, we’re selling to more enterprise customers and also like, yeah, on the visual delight, at least it’s also like a thing that’s, which comes down back to the connection always like always understand who you’re building for, who you’re serving with your product

EUREKA: design onboarding backwards

Jim (17:28): and start from there, like generate all the ideas from there. yeah. On the talk about AI, so one of the next topics is how is AI changing onboarding, right? So about onboarding, you have a really nice book, EUREKA, right? That’s how you call it.

Ramli (17:44): Do we have

Jim (17:46): In Greek, the way, think it’s a yeah, originally, it’s like a Greek word, heúreka. That’s why for me, it’s always confusing to.

Ramli (17:49): This. Yeah, it’s it’s an insight. Yeah, right.

Jim (17:55): And yeah, you have very, very interesting kind of frameworks to work with something that I. that I loved and actually one of the frameworks, it’s a bit outside of the AI but I wanted to start as a compliment on the book was if I remember correctly was you have to define point like what does success really look like and then you have to design from there like how you’re gonna get people there and.

Ramli (18:15): Yeah. Design it backwards.

Jim (18:18): Yeah, exactly. It sounds like a simple exercise, but for some reason I was like, Hey, this is really helping me think in a better way. I can come up with ideas. was like one of the frameworks that helped me kind of along some ideas. So thank you for that. And,

Ramli (18:29): Thank you. Appreciate it.

The four levels of AI in onboarding

Jim (18:31): the AI part, since you’re so much into onboarding, I’m really curious to hear from you. Yeah. The high level question is how is AI changing onboarding, practical patterns, for example, you’ve seen working in real life. And yeah, we can dive into deeper, like into onboarding and best practices and all, but yeah, let’s discuss more of the AI part first.

Ramli (18:49): Yeah, of course. Let’s talk about AI. It’s the topic talk about topic of the hour. It’s something I’ve been thinking a lot about because yeah, a lot of this a lot almost every product that we see, they’re adding AI to their product. and seeing I’m I’m seeing a pattern in terms of how AI is being used. I actually have broken it down to four levels. I’m noticing that there’s four levels to how AI is being used. in terms of increasing product adoption, accelerating product adoption, then product onboarding. It increases in complexity in terms of implementation, but I think it does increase how it can help the user kind of navigate and get activated faster. So the very first level is something I call AI tells. So AI is the one that’s, you know, there’s a chat bot there, and this is a classic chat bot that’s been around for many Many years, but now AI is kinda using and ingesting all of your knowledge base. user gets stuck and they chat with the chat bot. Hey, how do I do this? It would pull up the chat and you know, the knowledge base and tell you here’s what you need to do. Do this, go there, click on this. So just giving information. The next level after this is easy. You there’s tools that can ingest all your information. You can even use Claude or ChatGPT or whatever. and now they can chat with that and hopefully it doesn’t you know, hallucinate and make up features, which the early version of it has. So AI is information for product for product adoption. The next level after that is AI is actually guiding you. So in that case, when I say guide, it you ask it how do I do this? How do I set up an invoice? It would launch a tour. It would launch some kind of classic in app pattern or in app design. it could be a Like I said, a tour, it could be hotspots or whatever. So it’s highlighting specific areas of the spot that will guide you towards that specific goal that you have. and I’m seeing tools like Appcues and Userflow and Pendo, they have implemented something like this where it would connect what they call flows with the task or whatever the user is asking for. So it would kick start that specific flow that they have. So that one’s AI guiding the The next level after that is which where it gets really interesting. I’m seeing more products use AI to generate. So AI generating custom built templates for that specific user. So when a user comes in, think like Notion or Miro or things that is open-ended, where it can be used for a lot of different things. they would come in there and they would like describe us what you want. It’s like, I’m I’m a business owner of design company and I’m trying to manage my projects and it would build for you a custom built and generate for you a custom built template specifically to that. yeah like I said if it’s Miro hey I’m trying to do this it would generate that template for you. So you’re not picking a template which is the classic response or the traditional way pre-EAI were, I’m project manager, let me pick a template from this list of five. I’m trying to do this. This one it’ll actually generate it for you. So that’s a third level. That gets a little bit more complex because now you have to ha it to be connected with your product to create that template and requires more engineering connections and API calls or whatever technical terms, but it needs to create that template that would work for that specific product. The very last thing that I’m seeing more and it’s still at the very edge, and I don’t know how popular th this will become. is when AI is actually acting for the user. call it what you want an AI agent or a identification where the classic example I’ve seen is with Salesforce where their Agentforce would actually you respond to it, hey I need to update all of this records for this, assign it to somebody else is actually it’s not guiding you to that specific location. It’s not generating anything. It’s actually doing and acting on behalf of you that you approve it. I’m seeing something like QuickBooks where they’re still testing it out where create for me an invoice for this client and I need to send it out. So it’s actually starting to act on behalf of the user. That gets really interesting now because it might not learn the pattern the users might not learn the pattern inside of your product. Like how do I build an invoice again? Now it’s actually continuing to becomes dependent on the AI agent or that work the workflow or whatever to build that specific thing over and over again. And that get that gets interesting because it really does cut down the time and number of clicks to get them to that win, that aha moment. Because let’s say a project management tool is like creating a first task and sending it out. Rather than picking a template and doing all this stuff, he can generate for you, but then they can start acting on behalf of the user itself. So that’s how I’m seeing AI being used. I’m sure there’s more in the future where you can talk about AI video and you know, AI audio and all this stuff that’s on the side, but it is getting interesting on how it’s acting on behalf of users.

AI agents and avatars that do the work

Jim (24:23): Yeah, that one is an interesting thing that I didn’t have the chance to test with any of the products we’re working on, but yeah, it’s something I want to explore and see how we could run an experiment, see the reactions of people. Yeah, it’s more of like, just relax, sit back, and I’m gonna do it for you and you can observe. Like I’ve seen other tools, I don’t remember the name, where… they say we’re gonna, that they do customer success calls instead, like with AI avatars, which these are the ones that I’m skeptical, like, because we have the one is creating a new voice for me, and then you see the AI taking action. And I assume I haven’t used it, that you see some kind of a click on a button, so it’s not that suddenly the screen start changing too fast. So you kind of see the thing happening in front of you as if somebody is using, let’s say remote

Ramli (25:09): It’s taking over. Yeah.

Jim (25:11): control. Yeah, exactly. And then. The video avatars, I’m like, I get it, but it’s kind of a cringy to me. I don’t want to talk with an AI avatar. Maybe it’s

Ramli (25:17): Yeah. Same.

Jim (25:20): too early for that. Maybe that’s going to be the future. Who knows? Like, because every new technology, we need to feel more familiar and things that didn’t make sense in the past now feel the way to go and don’t feel strange anymore. So yeah, I like more the approach that you’re describing personally as a user. I haven’t tested which one works best. And I feel coming back to the connection of the delight, I feel maybe. both work just for different audiences. Maybe that will be the case. I have no idea,

Ramli (25:44): Hmm. That’s true.

Jim (25:46): but it feels like that. And when I’m trying to define if a pattern works or not, usually my answer is, it depends. Like you need to test

Ramli (25:54): It depends.

Jim (25:54): and see on the audience. But yeah, I also like how you’ve categorized the different levels of AI essentially, and how all this can coexist also, because people kind of attend.

Ramli (26:06): Yeah, they can.

Jim (26:07): I’ve seen on LinkedIn like people, yeah, usually LinkedIn, like debating on which one is best and who is killing who, but I’m like, yeah, maybe they, everything coexists depending on the product and the audience and so on. Yeah, it’s a very interesting one. And in terms of products for the AI does it think the one is the name, what are the some, Foldspace, yeah.

Ramli (26:27): Yeah. It’s called Foldspace. They do the generation and also the acting on behalf of the user. there I I did mention that it gets more complex as you move up this levels of using AI for product adoption. I some companies they build it in house. others they use third party tools like Foldspace to kinda help with that. There’s other ones that I haven’t really dug into but was the one that I saw. They talk about school they have that there.

Jim (26:59): Yeah, I’m really curious to explore more like on the, yeah, Foldspace. The integration part, that’s a very interesting one. Like it sounds like magical, but I’m like, okay, let’s see what it takes to actually integrate it. Yeah.

Ramli (27:12): To implement it, exactly. You need the engineering, you need the API, you need you need it to know your product, right? You need it to like integrate with your code base so that it can do the right calls, which is interesting.

Jim (27:24): Exactly. But I, as a UX kind of a product designer, can totally see that. And that’s why I want to dive into this and invest time. Like I can totally see value into having AI to facilitate some interactions as a way for people to learn. So instead of sewing, it’s like, let’s do it together. And we, we hire mentors, for example, at Love at First Try right to upskill our own designers. And one of the best things we do in the mentoring sessions is that we do it alongside with the mentor. Let’s say you want to build an AI native design system. Okay. Let’s start creating a design empty file. Let’s say together, just a random example. And that’s where we usually learn is when you do it with the other person. And I feel that’s a really interesting use case of AI because of course on LinkedIn, has been the debate of, like the new UI is just a chat. Like why would you even give people tables and all these things? But my current thinking is people need both. Like I need to be able, if I think about a scheduling software, I need to be able to see my calendar. I want to be to… be able tell like visually what is the calendar for the day, the colors, who is assigned to what appointment and not just interact through the chat. But for some other jobs, is a beautiful interface to interact with. That’s why going back to the layers of AI, it’s another example of like that they can coexist, which again comes down to the product and the audience. What are the patterns and interactions that actually add value to the users? Nice. And that’s what I love with products. like building products in general is that it always depends. Like you always need to learn your people and just find the right combination of all the beautiful ingredients that you have out there and make the right composition. Cool. Now on the next topic, what I have is real design stories with practical insights on activation. That’s kind of the theme. And yeah, you have a ton of experience. and you work with many companies and I’m curious to hear what are, let’s say the top two or top one design stories. By design, mean more kind of activation onboarding stories where you made some kind of change and you saw real impact. It could be also like some stories where it was the other way, where we’re like, this is gonna be a great idea, but it was a big failure because it’s very interesting. Just very quick note, don’t wanna interrupt you. It’s the… For me, one thing that we’re embracing more and more in the team is the whole idea of experimentation, which I realized as a designer, that’s my personal feeling, why I said about the failure stories as well, is that sometimes it’s scary to measure all the results because your idea may be, be, but, and it’s very easy to feel that, like I screwed up here. I should think of better ideas, but it’s the nature of growth that you try things and you learn through that. Actually, that’s why I’m saying that sometimes a great story is also a story of where you learn what you shouldn’t do. as well. yeah, whatever you want to serve, feel free.

An activation story: just pick one

Ramli (30:20): I’m happy to share maybe a company I worked with, I wouldn’t name it specifically without their permission, but it’s a tool for marketers and SEO and we already you can tell that’s their product was very complex. They can do a lot of things and I think that’s a challenge with a lot of B2B software companies where you know you can do fifty things or a hundred things and the challenge they had was that they were showing too many this is not new. Like whenever I come to a company and they are like, Hey, we need help and I’m sure this is your case too, where the founder comes in, it’s like we gotta get the users to do everything all at once and they come in, there’s a dashboard and it’s like, do this and that and that and there’s like ten things that they have to do. And this is this is not this is basic design product principles, right? That you know, the psychology around if there’s too many options, they stop. They stop. users would freeze up and they leave, they get overwhelmed. the mental overload, cognitive overloads too much. and what I work with them was talk going back to connections, talking to the user, understanding why were they coming in there for there in the product in the first place. you know, is it to write an SEO article? Is it to do research in their specific in their specific you know to crawl their website to figure out, you know, where are the vulnerabilities or opportunities? Is it this and that? So we I identified two or three jobs to be done or two or three core action that they’re trying to accomplish originally or the first one. and then just directed it out to the specific flow. So based on the response you know I would direct them to we directed them to specific parts of the product to accomplish what they wanted to do. and this unfortunately this is like, this is so simple, but it’s the simple things that make a difference. Right. And we had I don’t recall specifically how much it’s how much it’s in ta impacted the activation, but it was significant enough that they were quite happy with the work that I I did with them. So I think that’s just one thing that I’m trying to recall where like it’s those simple things of like Man, just pick just pick one. Right. It’s hard for founders to pick one. I think they wanna do it all. They wanna show

Jim (32:51): True. Yeah.

Ramli (32:52): it all. And so, kind of really focusing in the what to show to the to the user is super, super

How to find the jobs to be done

Jim (33:01): Yeah, I really love the thing in here is like, it does have to be rocket science sometimes, most times actually, like the simple things are the ones moving the needle. On this one, I’m curious on the jobs to be done, what are some ways that you… identify these because like a typical answer is like, yeah, you can look at the data, what people do, but I find this pretty chaotic. Like I’m like, okay, I just see events. How can I make, I can see pages that they visit and actions they take, but it’s super hard at least for me to analyze this. I’ve seen. using demo calls as a, like our customer onboarding calls as a

Ramli (33:35): Mm. Yeah.

Jim (33:37): source of knowledge is an interesting one. But I’m curious about your process on how do you find like the top three that say it to be done on this if you have a research framework.

Ramli (33:46): I you mentioned customer onboarding. If there are any customer onboarding costs, that’s what one of the things I dig into. Like why why did they jump on that? we signed up for the product in the first place. Why did they buy it? kind of trying to recall back to that first moment where they started looking for that product or that solution or you know, the specific pro solution to answer or respond to task they’re trying to do, a job, so to speak. And once they kind of figure that specific thing out, then kind of kind of direct them there specifically kinda I think the more it goes it’s hard to do it quantitatively and looking at pages. I think it’s it’s five to ten calls is usually enough to kind of identify who where they’re they want to start off from there.

Jim (34:41): Yeah, yeah, I hear you this. Yeah, and also like, yeah, and the whole thing with a qualitative versus quantitative, is because many times, yeah, you just need to go deeper, understand the motivation, like the emotional context, there are so many details, like calls, I find them beautiful, like there are so many small details you can get out of it, that yeah, it can really have an impact on your thinking, 100%. Okay, cool. Is there any other story you want to share? So we go maybe to the next question. How do you

Ramli (35:15): Happy to go the next one. Yeah.

Ramli’s AI workflows

Jim (35:17): Okay. Yeah, it’s high level questions that I always love. asking guests because it’s people that I find great at what they do. And I personally trust them. And I am really curious, like, how do you do, how do you, AI is the exact question, which means like, how do you incorporate AI into your work? And that could be from your consulting, because I think you have the inputting, the activation screens, you have your community, you have a newsletter. Yeah, you have a lot of things happening. So I’m curious, what are, say that the top three use case where you’re like, I love AI. It makes me feel great and it’s actually helping me do something and I’m happy that I’m paying for it.

Ramli (35:57): I think there’s a few ones. It’s gonna actually tie to the other question around software. you know, I I do a lot of calls, calls to product leaders, hosting peer groups for product leaders, calls with clients. And one of the things that I when it came out was like, thank goodness was I before I was I’m using a tool called Granola to record my calls. And I would just copy and paste all the calls one at a time. But one of the things that it’s like, man, this changed everything was having an MCP from Granola to Claude. and then having like this ongoing daily agent picking up my calls. Like what you know, what meetings do I have coming up and what context did I do I have with them? And kind of pulling all that information as to like what are next steps. But I think so that’s one of them, pulling them in and having meeting summary. But the second cool kind of AI thing that I’ve started pulling is with all this peer group sessions that I have and calls with product leaders. I have another agent that kind of pulls that and suggests on a weekly basis like what are LinkedIn posts I should be kind of writing about or content or articles I should be thinking about or writing about or hot takes. one that came up recently through a discussion with a product leader was around this bubble that we’re in and you know AI’s getting more and more expensive. Like that some people blow through their whole tokens in several in two hours because they were just vibe coding. So the qu the real question

Jim (37:35): Right, yeah.

Ramli (37:36): the real question is like when it when is when is it better to hire a human? Which is the opposite of what we’re all talking about now. Is it when is it cheaper, faster, better? and so I think that was an interesting insight that I turned into LinkedIn posts and a blog article as well. I think the other thing is just I’ve create connected my Claude. It’s yeah all my things are connected with Claude, connected with ClickUp. So anytime there’s an update with a call with a client or a peer pod, some you know peer group session or a community member, you would update all that over to ClickUp with their specific kind of account so that it’s up to date with any kind of information that I have. So if I had to break it down to three things that I was like, shoot, this is this is helpful. It’s first of all the calendar summary, meeting summaries that I get before and during and after for next steps. Second is around pulling out content that I that I get from discussions and insights. that comes from Granola again and then goes to Claude and Claude would suggest a few things that I would post on during that week. And the last thing I’m just updating, making sure that my ClickUp Task Manager is up to date with different information about my clients, my my customers, my community members and anything else that I need.

Jim (39:01): That’s cool. Yeah. And I love like the example of the transcript, the copy paste. I think it just gets connected to the previous example of like the activation of like simple things can, yeah, sometimes it’s just the simple things. It’s about just like super manual tasks. It doesn’t always have to be this magic of like replacing something or coming up with a insanely kind of a smart insights. Sometimes just about the boring tasks you want to Yeah. Get out of the way actually. Yeah, and Granola, I’ve been using Fathom for a long time and I started using Granola as well and I love it. Indeed, I get it now. Indeed, the action items, the summary is great. They’re great. There is really high quality of the output that AI actually generates. And yeah, I also love like the whole product experience. On the…

Ramli’s favorite software

Jim (39:52): Yeah, since we were short in time, like I have one last question that I wanted to ask you and it’s about what’s your favorite SaaS. Also,

Ramli (40:00): Mm, yeah.

Jim (40:01): by the way, it doesn’t have to be SaaS. mean, we’re all about SaaS here, but it could be like B2C app or something else. For me, it’s more like some kind of a technology product, a software piece of software you’re using that comes to your mind. And why does it come to your mind is the top one.

Ramli (40:17): Yeah, I would I already cheated a little bit and mentioned Granola. I it I I don’t think I can survive without it anymore. that’s a funny thing to say. Just just so wrong. No, it’s I don’t just use it for my clients or peer group sessions for my community members. I really like having the mobile app for it. So if I’m in like a meeting with a doctor, I would have Granola in the background or I just recently had a parent teacher meeting. for my three year old kid, and I was recording that so that I can have next steps and actions. and so I have a bad memory. So I mean, maybe my memory’s not that bad, but this really helps me kind of try to pull it all together in one place. It’s made my life easier in terms of like all the things that I have to kind of remember. It’s like, what do we say here at this time? It would be a scary world where it records everything. And so like what did I say to my wife like a month ago? That would that it would be first of all creepy, but it would like shoot, that would be helpful. ’Cause like if we’re having a discussion, like did I did you tell me to do this? I don’t remember and just bring up Granola and it would be that. But I think that’s one that it’s really made my life easier.

Jim (41:41): That’s a good one. Yeah. It’s very interesting with Granola that it just keeps coming up. I’m like, it’s just a note taker. What’s big deal with it? And I had so many guests, yeah, so many, like from the guests that I had, like a big percentage actually said Granola. And it’s very, very interesting that it’s the simplest product from old. It’s literally just AI notes. And I think there are like a hundred tools out there, but it’s the one winning again, again and again. Yeah, also side comment, a bit of topic, but on the thing that you mentioned of like recording everything in the day, that will be, indeed it will be creepy, but I really

Ramli (42:14): You would be

Jim (42:17): like what you said there. It could be also interesting because one thing that I have with myself, for example, is like… saying things with a specific tone, but yeah, that wouldn’t be on the transcript, it would be more on the voice. Where like sometimes people may say that it comes more, I don’t know, like aggressive or like the tone of voice where you don’t really understand it when you’re saying something. that’ll be an interesting actually AI tool in terms of transcription actually to help you communicate better based on kind of the style and everything.

Ramli (42:42): That’s true. We could analyze your voice. Yeah. That’ll be making sense.

Jim (42:44): Yeah, that would have been interesting. But yeah, cool. Thanks. Yeah, that was it for the episode. That’s all that I have for now. Thanks a lot for coming today. Really nice insights and really nice high level frameworks to think with. Honestly, that’s I feel like a thing about the takeaways from the episode. There’s some practical frameworks and mental models the people that are going to watch can use. So yeah, thanks for sharing everything. It was really nice having you here.

Ramli (43:14): Awesome. Thank you so much. Thanks, Jim, for having me.