Zero to Launch Panel NYC 2019



Transcript:


Alright, so today we just got a lot of information from where we were a couple years ago to where we are now to where we're going some of the best tools in the industry some of the best tactics to deliver these business Bots for your clients or for your business. And now we're really going to focus on what you need to do to go from idea to finish product. We're not finished product, but just idea to a minimum viable product that's out there that people are talking with and you're getting data to improve.


So that's where we're really going to focus on right now. And right now we have a really good panel. Again. My name is Joseph on the co-founder of smart Loop where a chatbot and live chat platform really easy to build Bots to automate your marketing and customer service and I'll go ahead and pass it on and we'll have introduction. So, my name is Sansa Dum. I'm one of the founders of robocopy and the conversational Academy and what we do is really recognize develop and promote the role of the conversation designer. So we train and certify conversation designers and with more mics.


Yes, so we help with that and we help companies really develop conversation design teams and have them work in such a way that they can really scale out through the organization. I want my name is a bit. I'm the founder of bots Mark we are conversation design and prototyping tool we enable teams small and large teams to actually design the conversation flows understand what the experience would be on 11 different channels from Facebook to wise to slack Microsoft teams Enterprise Bots and web and mobile and we help you do that from design prototyping usually testing and the whole yard and then we have a developer API that enables your Dev teams to take what your design and prototype. Take it all the way through number of platforms like razib Microsoft bot framework IBM Watson dataflow or your own. Hi, I'm Joe Austin. I'm the CEO and founder of weight and Wendy.


We apply chat. Bots and AI chatbots for the recruiting space. So we have two core products one that enables us to screen inbound applicants that apply for jobs specifically for the role be able to understand and comprehend the response is and then be able to make recommendations back to the recruiting team on how to optimize their workflow and on the other side for roles that don't receive a lot of applicants. We focus on the sourcing side of things which is engaging and presenting people information about job opportunities helping them learn about the roles through chat and if they're interested getting them back over to the hiring team, we're venture-backed startup. We've you know, we've raised from some great VCS and we're based in New York were 25 people and we're live with Enterprise customers big Banks Fortune 100 companies. Is of this chatbot Revolution, so some of the things that we learned about building chatbots, I think were some hard lessons that I think people have now put a lot of infrastructure in place to be able to assist so hopefully we can share some light on that go gingers.


My name is Dmitry.

I have marketing agency called Target choice. We are focused on running Facebook ads and building Facebook Chat Bots Facebook messenger chat Bots and there is something that we do that is a bit different from what most people who are doing Facebook Chat. Bots do which is we build self-hosted on premise chatbots which opens a lot of interesting opportunities that are much harder to do. If you do not control the platform. You cannot have your own database. You cannot inject code we can Build a really deep really rich applications which look like mobile native apps, but they in a messenger and they're integrating this kind of functionality into the conversation flow. Lots of opportunities. Not many examples of this that you can find out there right now.


Hey, my name is John Wolfe. One of the lead developers on Bots. I that's our chat bot platform. I'm here for a send them digital. It allows you to quickly get up to speed with your knowledge base data in bed anywhere and rapidly scale out a chatbot. One of the couple of the other things. I do are quickly proof of concept different ideas getting them in front of people running analytics on those proof of Concepts and seeing what sticks and what doesn't and quickly making decisions from that. So, hopefully I can share some of that with you today.


Mason Levy co-founder and CEO of civil we simplify the AI training process ultimately helping companies take out the trash through guided data tagging exercises that they can either do in-house or Outsource to focus on two main competencies. First of all speed to launch and then closing the feedback loop. So these systems get smarter quicker. Okay, great. So we're talking about use cases.


So when you think of an Enterprise, what are the best use cases that an Enterprise can focus on how should they be thinking about chatbots AI invoice and what case studies have you really seen shine in an Enterprise? Well, I think the obvious one is still customer service, right? And that's where we see the most success and what our companies really get the most value but you see them spinning off at different department’s right where we use persuasive dialogues to just get people to buy products or book meetings or download papers. I think what Enterprises really need to be thinking about is how we going to sort of use AI to communicate with people and how we sort of going to design a boat's Department over time. So I think you sort of see now different use case.


Every Department can have his own little use case, but how you're going to really create a department that's going to run that. And make people feel comfortable around that. So now we see it's a lot really like an engineering problem. Most of the time we feel we need to have much more emphasis on sort of dialogue design conversation design, but also understand the psychology and everything that goes into that right? So the really use cases there everyone under super exciting but a lot of companies aren't getting to that point that because they still see it as an engineering problem. But if they invest more in the design part and understanding the psychology, then they'll actually get to execute on the other use cases. So that's sort of our take on that as a form our end. We see a lot of our Enterprise customers starting from either customer surveys or internal sales. Some of them are internal facing pot.


Some of them external facing we have some customers were doing if entertainment so they're bringing games and other Experts of their brand engaging with their customers and the idea is that there are trying to get a dialogue going with their customers first. Learning from what their customers are asking for and then even deep in the string in the second kick at the can and then maybe seeing what else can do can they do so, you know form wise, you know, it's all about don't go transaction right away try to provide value first and then see what they're asking for and then add more value in transaction more and then on text as well this customer service and then other people coming in. For us. I think what we looked for was use cases that really showed the value of where chatbots perform. So we look for areas where there was high volume of interactions that when we look deeper had Repetitive work flows and repetitive types of interactions. So I don't think it's as much about like, you know, I see a lot of people try to squeeze chat Bots into a wide range of use cases, but I think when you really looked at when you really kind of chart try to like break down what the purpose of that chat bot is, you know, it's to present information gather information.


Hopefully synthesize that information and then ultimately make some type of recommendation based on that information. So, you know things like scenarios that I think are really interesting, you know, you look at real estate, you know, I've been I've been looking for an apartment lately and it's a very complicated process for me, but at the end of the day, we have a chatbot help Net helped me to navigate the information that I need to give similar to the tax one.


That was read that Ayah that we saw earlier. I thought that was a great workflow for us with recruiting. I was a recruiter I spent five years in recruiting I felt that the types of conversations that I had no matter how many tools and systems we had in place. The conversations that I had to have where the most would continue to constrain my ability to scale my job and my performance. So when I was able to really break down what I was talking about and then look at how we can offload some of those interactions so I can have more human interactions with people instead of racing through chats. That's what I looked for. I looked for areas where repetitive workflows repetitive conversations clearly. Do you know there was very like clear intent and data being exchanged so that we can make informative decisions based on that and it's kind of how I've looked for use cases.


I see the best use case as the simplest sell more stuff make more money. That's it. Right. So how does it break out? This means you have to look at what is acquisition cost of getting somebody into the messenger. What kind of drip campaigns can you get in front of them to get them taken action? Whether it be a just generating leads so they have to be sold off messenger or you're doing e-commerce but truly comes down to your marketing metrics. How much does it cost to acquire a user for the bot? What kind of offers you're putting in front of them how you're converting them and how much better? This is versus other marketing channels. Now, how do you do this? It's also pretty simple you look at your existing sales scripts you look at the existing questions. You ask a prospects and you turn it into the bottle workflow.


You put it in front of lots of people you optimize this you keep dialing in on this and you end up with Positive Roi so I just heard real estate being mentioned. This is one of areas of our focus and this is just amazing how much opportunity there is in pretty much every industry to take existing sales process existing Prospect qualification and sales process and turning this into a structured process of happens inside of the messenger and also integrates other platform the new integrated with Facebook ads you get it in front of pretty much everybody in the world. So that's where I see the biggest use case. Yeah, I think some of the things that we've seen are speaking a little bit more generally anything that's a pain it takes time and Manpower and can it be automated so when it fits those categories you can have a chatbot solve those kind of problems and quickly and I think of several people have mentioned return on investment. But if you're saving man hours you're saving money. So yeah, I think you know AI automation chatbots is kind of allowing us to rethink just about every business process. There is from an Enterprise perspective. I think some of the internal business processes are the most exciting for automation.


So you have HR recruiting all the way to payroll had a lot on board a customer. Our employee off board that employee as well as project management and assisting through that so that if you have a remote teams or you know people that are working different hours you kind of have this man in the middle that's helping communicate at all times. I think those are some of the more intriguing use cases for Enterprise. Okay, great. So now that we know the use cases were coming into lightning around here. So we're going to just say if you could tell us like what is your number one tip from getting to get that mvp out the door so user can test and you can see how people are using it and then if you have any tips on like how to launch a bot Right. Yeah. So when you start out don't focus on technology.


So what you want to do is really figure out sort of the from the users perspective. What do they want to achieve? But also how did they feel when they want to achieve that? So what are the things ayat? He's and motivations and sort of designed for that and we use roleplay for that to get a good feel of what the dialogue should be. And once you have that structure, that's something that you can validate real quickly through Wizard of Oz testing and that you can implement it and actually save yourself a lot of time in the design process so you can have a chat bot that way at the end of the day instead of just diving into the technology straight away. So you'll save yourself a lot of headaches that way. And just to build on top of that right? We obviously our customers do this every single day on our platform. They are starting to explore an idea of a chat pod. They come on platform. They're going to prototype it and that's what we provide them the tools for and our job and our submission is to give them that ability to get a prototype in front of their customers without writing a single line of code without even going to the platform itself.


So on our platform, you can build the entire conversation experience on Facebook Messenger for example, and send a link to somebody that doesn't require involved in physical machine. It does not require them to do anything like fancy like that, but you can run focus groups get feedback get input just from your team stakeholders and then outside of it. Once you get that feedback. You are already 10x better in terms of understanding is this worth pursuing and putting all the engineering resources behind it.


So we usually look at them and say take away all your a IML and all the fancy stuff out first just write dialogues first was Create a flow diagram. You will understand your customers Journey focus on a use case that you want. Give them the task and say hey, here's a sample chat board composition. Try to order pizza if you're building a piece about right and see how they struggle and what's their The Experience look like and you will learn a lot.


Yeah, I guess this would also a little bit go back to what I was saying. Last time is like I think a great way to validate it assumption or a test with a chatbot is to try to find an existing try to find an existing flow of users that you can kind of interject yourself with. So, for example, when we got started, you know, our first idea was you know, I was a recruiter and I said, I want to go fine people. Let's take companies that are hard to fill jobs and try to find people to place them. But you know for that the challenge you have to face is not is building a database of people first before you even start engaging High volumes of people. So when we shifted our so we said, how can we just interject our self in the workflow to be able to just engage in and try to really get At people through the pipeline so they can test the experience. We moved it to the post apply process so that anyone that so we don't have to focus on acquisition and chat. We could really just focus on the chat side.


So I would say is focus is one of them focus on really testing your core assumptions and the second thing is identify clear kpis. So what is the actual metric and there's a lot of fluffy metrics in this space. So let's be like there let's be real but they're really understand. What is the key metric that validates that chatbot in this space and in this situation and this scenario makes sense. Is it start to completion? Is it being able to demonstrate trust? Is it executing the task? Is it conversion metrics? Whatever it? There's a golden metric there and make sure you see you move past any of the nonsense metrics and get to the one that matters most and then focus on limiting anything else that can distract from validating whether this works based on that metric. Well, I wanted to talk about something very similar. I would say focus on solving problems that are worth solving and find the low-hanging fruit. One of the challenges.


I see with a chatbot space overall. They're just so many shiny things to distract you. It's so easy to turn a chat but project into sort of an exploratory science project and it's cool. But the question is how much impact is it going to make on your business on your goals where we try to focus is sales revenue marketing.


It's very easy to make a case. How many more customers we signed up how much more stuff is sold them? And is this the right thing and how can use it? Is it we already have to reach those people? How do we Leverage The Creative assets find something that's going to give you a quick win that's going to get people in your organization on board with chatbots and asking for more another hand. Sometimes people just want to experiment just create something that is completely, you know, interesting exciting but not necessarily practical but if you do something like this, it might be challenging to get your organization to keep supporting the project find the low-hanging fruit and pick it. Yes, the speaking to the first part which is getting it in front of your users some things you can do especially at the Enterprise level is partner with your end users and work together to come up with something that you know is going to work for them have a team ready to scale it out at least in POC mode so that you can get it in front of who's going to be your end users as quick as possible and iterate on it.


Of course, you'll need your analytics in place to do that. But you knew that ahead of time so it's going to work and from there actually launching it you can also work with your end-users you develop some Buzz create some hype, you know, you have a demo you have your sales pitch and after that it's all about the infrastructure you've set in place and set a go live date and go also I'm trying not to repeat answers so, you know talk about tips. Tricks it's about having the right Tools in place.


You know what I look for in a bob building platform or typically for things, you know, how easy is it to implement and get up and running because real live customer feedback is better than anything, you know does the platform is it flexible enough for me to experiment and optimize and real-time. Can I train and tune my natural language processing or understanding algorithms really quickly and then ultimately because we're putting this out into front of real world people really quickly, you know, what's the escalation process look like to push to a human and making sure that that feedback loop for human to human interaction is actually retraining and making sure that the body is getting smarter as well, too. All right, cool. Great tips. I mean, I think the key thing is here is just conceptualize your idea figure out what you want to build and build it don't get paralyzed by research and spend months and months and months and never do anything.

So do simple things good.

© 2017 by Chatbots Conferene TM.