 I'm not sure how to talk after that intro. Okay, so I think none of it, I think Abilash, you know, really good sharing, Leon really good sharing. I think they touched on a lot of really interesting things that, where are we going to use the mic right now? It's fine. I think they touched on a lot of interesting things from, you know, a builder's perspective. I think from my side, you know, so Alex and those guys, they touched me with sharing a little bit more of the serious stuff. So I think I'll talk more about our experience and also like, you know, share a little bit more about, you know, how do enterprises, how can, you know, like bigger companies think about bots as well? Because I think that's also one of the things that I think many of you might be interested in, you know, where are the opportunities, what are some of the implementation things, what are some of the difficulties and all this kind of things when it comes to pushing bots forward for enterprises. So a little bit of a key reply, I think we are trying to build one of the better companies in this region as an AI chatbot platform for enterprises and government. So myself, I used to work at, you know, Twitter as well as a SingTel. That was before my co-founders and I started a company. So I think Arbilash and Max, you know, I think 2014, they won the Facebook hackathon for the first time and Max won it a second time again in 2015. So that shows that he's extremely smart. So I chose to work with the right people. So, you know, so we have been hustling together for the last like two years now. We started in Singapore. We started work with collaboration products, chat automation, chat widget, summarization product, you know, and all this kind of good stuff. We recently did three accelerators in the US, Angel Pad, plug-and-play travel and hospitality as well as blue startups. So if you guys have questions about US visas, anything, ask me offline. I'll be happy to help. For us, we actually made a decision about three weeks ago. So we actually came down for, you know, just for this event. But also because we realized that there's a lot of opportunity in Southeast Asia that a lot of the US companies, the bigger ones, are not looking at. So, you know, we work with partners like Facebook, we work with partners like Microsoft, you know, supported by those guys and really trying to do more in this region with more and more companies. Yeah. So you might have heard of this before, right? A lot of, you know, your colleagues, your directors, your CEOs or your friends are like, I want a bot, right? Okay, so they've seen Buzz Uncle, right? They've tried Buzz Uncle. They're like, oh my God, I want a bot. So, do you just launch a bot? Do you just want to be the number one company to launch a bot? Or do you just want like AI now, right? These are some of the common things that we've been hearing when we talk to companies, which is not a bad thing to start with, right? Because it's like something that, it's a good starting point where, okay, now you care about this topic, right, which is good. But I think the reality is that when you think about, you know, this entire thing, there's a lot more consideration that you have to think about, especially for example, if you're representing a bigger company. So, one of the first, most important things that we realized that we always tell people when I have meetings with people and stuff is really to define what the objectives are for, you know, why you're doing this, right? Because for some people, it might be innovation, right? You're trying to do something like you're the first in the world, best in class and so on. For some people, it might be, you know, let's just help our customers get immediate support. Meaning, for example, someone is asking a question and you want to give them a response immediately to get the check mark on the Facebook page, right? That's great. Or it could be lowering costs, right? Because at the end of the day, businesses drive on bottom line, you know, reduction, which is great. Or it could be about differentiated service, right? For example, right now, your business is focused on just selling maybe drinks to people. Maybe if a bot, you can actually offer them, you know, drinks in their location and so on. So, you know, all these are good ways of thinking about it. Another way is maybe personal touch, right? Like how bus uncle designed a personality. You could potentially design a personality for your brand as well, you know, maybe ask Jamie, you know, for the government services and so on. Or it could be an AI, having an AI strategy, a broader AI strategy, right? Right now, I think one of the strongest bus words in the media that everyone talks about is AI. Like you can't really read Business Insider or Venture Beat without looking at the word AI. So, it's important to know that, yes, okay? AI strategy is the legitimate reason for why people are doing bots today. And maybe last but not least, increase sales, right? At the end of the day, like for small businesses, like, you know, Alex and Leon mentioned, a lot of the small businesses are caring about increased sales, right? Because at the end of the day, they can't just drive on like, you know, buy and sell itself. So, for whatever reasons you guys are thinking about for bots, you know, for companies, these are all some of the potential reasons that will be interesting. It's not as if like, so what I'm saying here is that there's no right or wrong answer. It's just that you have to know what those reasons are and you know, really start thinking about that as you start, you know, looking at your initiatives. The next thing is identify internal stakeholders, right? Some teams, like for example, the marketing team, they say, oh, I tried this really cool thing. Let's now go and build a bot. Then they talk to operations and operations like, hey, hey, no, no, no. I'm not going to let you throw all the support tickets or like, you know, integration onto my plate, right? So, identifying the right stakeholders doing companies is very important. You know, getting the CIO involved if you have to, please don't, if you have to, marketing teams and so on. So, getting the right buy into the same room, actually helps in making things a much more streamlined process. Or for example, if it's just within one team, that's even better, right? Because things will move faster. Then when it comes to like, you know, maybe an internal sort of thinking about it or when you will talk to customers or talk to vendors and so on, you want to map your conversation flow. This is actually a real one. So, I actually made it really, really small so that you can't actually see it. Yeah, but essentially how a conversation flow can look like, right? You want to identify all the age cases and you know, of course, at the end of the day, even though you identify all the age cases, users are going to ask you anything, right? So, this is just a start, right? This is not the end, this is just a start. And then of course, this NLP and machine learning thing that you heard about, this is really complicated. And maybe it boils down to two sentences. It's like for example, for a telco, how do I change my IP address on the router? How do I edit my IP conflict on the console? If I give this to a three-year-old to look at a text or six-year-old or seven-year-old or whatever, he or she won't be able to understand this. But with the full context of IT training and so on, you will be able to understand that, yeah, it might mean the same thing, you know, in this context. So, because of that, you know, and NLP is a really hard problem, like, you know, it's no joke, it's a really hard problem. And, you know, right now in the current state of research, there are limitations, you know, for example, yes, the media talked about a lot of things, like, you know, hyping up, you know, the possibilities and so on. But fundamentally speaking, we are still at a point where we're trying to use, trying to understand language better. A lot of the, for example, the Stanford Q&A data set, it's called Squat, if you guys are interested in the research side of things. They publish a recent data set of about 100,000 Q&A samples from some Wikipedia and all these different public articles. And you can actually sort of run experiments with it. So Singapore Management University actually published a paper for that. And the result was pretty good. They were the first one to publish a result for that, actually. So, of course, now Salesforce, Microsoft, and all these guys have published better results. But, you know, I think it's interesting how the research community is really, you know, trying to rally behind this and trying to solve this at scale. And of course, on the production side, there are APIs available. Some of them are not enterprise-ready. Some of them are. So you also need to identify what are those that you can work with and understand from your perspective whether you want to use them at all. And then I think one thing that you want to think about is also integrations and channels. So, integrated channels have two sides to it. The first is actually the front end, right? For example, when enterprises launch a bot, they will think, hey, I want to get a Facebook Messenger, web, a mobile app, line, WeChat, and you know, a ton of other ones. They want everything. Great. Is there some that your customers are more on? So I think a good way of thinking about it would be where is your customers currently on, right? Are they on WhatsApp? Are they on Facebook Messenger? So of course, WhatsApp has not launched a bot platform yet. So, you know, we are not able to integrate with that. But, you know, you can look at other platforms that people are using and really focus on those that make sense for your business. Another thing is really backend systems. So if you think about backend systems, you might have existing CRMs, you might have existing call centers, and all these things set up. You don't want to waste this infrastructure, right? And it doesn't make sense. So you need to think about how can you integrate with that and make an informed decision. You know, if you want it to be isolated from that, great. But if you want it to be combined, then you need to think more about all these kind of things. And then distribution. So distribution, you know, is, I'll talk a little bit more about it later. But there are many ways and it really is about getting in front of the people, right, whether it's about PR, whether it's about ads and so on. Analytics and engagement. This is very important. So Facebook launched their Facebook analytics for bots. At the same time, we also see a lot of bot companies being around, for example, analytics for bots, and there's a reason why they exist. So they exist because I think some of the tools that people need to identify the conversations, for example, that people talk to about, where they fall off and all these kinds of things, are not currently supported yet or not currently built by the Facebook team, or you are deploying across multiple platforms, right? So you want the combined analytics across different things. So all these practical considerations do come in and you know, you need to make sure that you have that. So, bots for your company. How do you do it? You can do it in-house, right? There are a lot of companies who do it in-house, like maybe, for example, have an internal hackathon. Hey guys, let's use chat fuel or bot framework and so on, and let's do something, which is good. And then, but the only problem with that is that as with a lot of internal hackathons go, the projects don't last past the hackathon, right? Which is kind of sad. So you want it to sort of have a strategy, okay, if we do a hackathon and there's some really interesting projects, let's try to move it out and commercialize it. But it also has problems itself because projects, as we know, build doing hackathons are a little bit more shaky. They're not really meant for, you know, kinds of people to use. So that's also one of the problems. You can also do POC with maybe selected users, right? Have a small focus group, launch it to them. These are all things that you can do. I think importantly when you do it, after you launch, you want to really understand how users are using it, how you're segmenting the user groups, and then of course test more things and optimize further. And of course, use this as a way to inform your AI strategy or whatever that is. Or you can look at a vendor, right? So I mean, I'm biased because I come from that side of the business. So the thing is for vendors, I think if you talk about, if I were to evaluate vendors today, I'll be evaluating vendors on a couple of things. Pricing is important, but it's not the most important, right? Pricing is something that, you know, it should be focused on results. If the pricing makes sense, it's results oriented, that makes sense. I think portfolio of customers is an important criteria because you want to know that these companies have done things before, right? So I think that's also from our perspective we have to think about how can we build a better portfolio of customers going forward. Another thing is I think timeline. So some companies can wait one year to launch something. Some companies cannot. So depending on timeline, whether it's three months, whether it's two months, whether it's six months, it also will help you to understand what is the scope that you really want to scope this into. NLP and ML, I think is an important thing to think about when evaluating vendors as well. Do they do it in-house? Do they use existing platforms? Do they integrate with multiple? Or how do they think about it, right? And, you know, it shouldn't really be like a black box, you know, like we don't tell you what we do kind of situation because that's kind of dangerous, right? And you don't really want to be buying into that. Management is very important because when you launch a bot, if you have the team, your content team, your social team and your support team using it, you can't have like, you know, like one person managing like 10,000 messages every single day, right? You need to have a management dashboard and you need to have analytics and audit and all this kind of thing. So, these are some of the things that we think that we see is important for customers, yeah. And the pitfalls, right? I think it's easy to launch a bot for fun, right? It's easy, like you can do it over like a day. But then, what's next, right? I think if 35,000 developers develop bots for fun, then we have a situation like what we have today which is like, I think there's a lot of bots out there that's not really well thought through and I think we might even be guilty of that ourselves. So, I think that's something that we want to sort of help to maybe like, you know, change going forward to focus more on quality, focus more on results, focus less on a hype and focus more on what can we do meaningfully. Age cases, I think, you know, is always going to be there, right? So, when you launch a bot, you know, you're going to get people asking questions in different ways, all kinds of ways. So, you want to make sure that you iron out some of those age cases and continuously optimize. So, the bot doesn't just stop after you launch, it actually continues, right? Like what Abilash has seen, day one to 30, the architecture looks very, very different. So, when you guys do your own personal projects as well, you'll see that happening. No improvements, feature creep, these are all some of the things that, you know, comes with all these like app development projects and websites as well. And I think, that's one of the least, I think important integrations are always like, you know, sort of not looked at because sometimes we just want to launch things fast. Yeah. So, testing. What you've seen is that these are three main ways of doing testing. Internal testing between you and, you know, your team members would be really, really good. And then actually about 60, 70% of issues can be sort of taken out from there. And then, you know, when you travel with customers, then you get a whole ton of new use cases, new tons of problems and everything. So, then you of course, you smooth out all those things. And then when you start scaling, you realize, hey, maybe I don't want to do all these things myself. I can do maybe more automated stuff. So, you start writing scripts, you start writing programs on the backend. So, these are all the good ways to think about it for testing. For launching, I think there's four broad areas where you can do, you know, I think a couple of like things that maybe I didn't mention if you have partnerships and so on. But broadly speaking, using your existing assets on the web, on the app and so on is really, really good. Email list, if you have. And I think right now there are some Facebook or Google ads that you can actually run. So, Facebook, you actually can run this ad format called send message. So, it appears in people's timelines and essentially people, if they're interested, can click on the ad, send message and then get started with the bot. So, that's actually one of the ad formats that's available. I'm not sure what other formats would be available for, you know, for, for bots going forward. But I think you'll be more and more interesting. We also see re-engagement tests from Facebook site as well. So, for example, let's say within a messenger conversation if someone has talked to the bot before, you can trigger ad unit actually re-engages the user. So, that's pretty interesting. PR and content is always helpful, right? I mean, if the bot solves a fundamental daily use case, you know, or like event-based use case and there's a lot of interest around that, then I think that's always going to drive a lot of usage for the bot. And I think last but not least, I think this one is something that, I think as an industry, as a growing nascent field, everyone is still figuring it out as we go along. But I think going forward, there'll be three main types of metrics that people will care about. The first one is actually engagement metrics, like core engagement metrics, how many messages are being sent, right? How are people talking to the bot? Are they talking to it every day? Are they talking to it every week? Are they talking to it every month, right? And then you see dialogue. Dialogue meaning how are people communicating with the bot, right? The actual conversations itself, are they using the buttons? Are they typing stuff? What are they typing? Are they typing long messages? Are they typing short messages and so on? So these are all the things that you want to be able to track and be able to understand. And this is also useful information for the company itself. And then use case. So use case-wise, sometimes it surprises us when we launch how people use bots. So for example, to give you an analogy, when we did the ASCAC bot with the ASCAC team, one of our primary objectives was, let's allow people to access jokes and funny things more easily on Messenger. It turns out that that's only one of the use cases that people use it for. And we were surprised that a lot of people actually use the bot as a way to send images and submissions to the ASCAC team. So we have to literally basically build a new feature, which is essentially allowing the escalation of the images through email or through the backend to the team itself. So that's actually some of the things that we've discovered ourselves, that sometimes there's going to be all these kind of unanticipated kind of things that happens as a result of you launching a bot experience and letting users try it. Okay, so now, yes, I hope that you guys do want a bot after hearing me ball through this entire thing. Hopefully you guys do want a bot. I think the main takeaway I know that I wanted to share is really just think of implementation as stages, as with anything. If phase one today is launching on Messenger, maybe phase two today, tomorrow could be launching on Messenger and with integrations with backend systems and so on. Identifying right partners is important, whether it's Microsoft, whether it's Facebook, whether it's Amazon or any of these companies or like Vendors and so on, or your internal stakeholders. These are all very, very important things to figure out. And then you can have a proper plan to move forward. And then tracking relevant metrics, right? Every time you start a project, don't just focus on innovation. Innovation is good, but try to think about the metrics that you can drive for the business. And if not, you won't be a sustainable project, right? It's not going to be sustainable as an effort going forward. And we also don't want to see efforts fail, and that's very bad for the entire industry, yeah. So this is a bit of my plug. So we have worked with the government of Singapore, Ministry of Communications, Tech in Asia, Ninja Van, ASCAD, and of course more customers. Some of them I cannot disclose. And then this is actually some of how we think about it. Not just about deployments or Messenger on web, but also a full-featured management dashboard for customers to be able to use, for content teams to be able to use, manage all this information. And of course, the feature list kind of goes every single day. Enterprise software, that's how it works. So yes, we are hiring very aggressively right now. So if you guys are interested in learning with us in this journey, shoot me an email. Let me know. If not, thank you guys very much. Anyone got any questions for Spencer? We're learning questions. We can take one before the panel session. I'll take it for the okay hand. Later you try to post it on us or on the, compare the speakers. I'll be sure. My hands up. My hands up. My hands up. Okay, not right. So ladies and gentlemen, we will have a very quick.