Revolutionising Customer Service using Chatbots with Paul Shepherd

Host: Marco Oliver, Client Success Director at We Build Bots

Guest: Julian Harris, Head of Technology Research at CognitionX

Marco Oliver: Hi Paul and welcome to The Botcast.

Paul Shepherd: Hello.

MO: So obviously it's quite strange interviewing you because you're my CEO but we'll give it a good go and I'll give you the same grilling that I like to give the guests generally. So, firstly we've had a really good year, you particularly of had a cracking year. We won the prestigious Sir Michael Moritz Awards at We Build Bots, we’ve raised over £700,000 on our first crowdfunding campaign and recently we've been accepted onto the Oracle Startup Cloud Accelerator Program, so it's all really positive. Before we start to go into more detail on that, do you want to just give the listeners a bit of background about yourself and where we've come from and where we are now?

PS: Yeah, sure. I worked in Digital Communications for Financial Services, I then left that world and set up a digital agency.  As part of that, I spent about two years on a big contract in the state's, going back and forth every couple of months. When that contract ended, I settled back in the UK and started building the team to about 9 or 10 of us and the agency almost became an incubator for product ideas as much as a service agency and a lot of what we did was managing the social media feeds for customer service departments and that really spawned the idea of what is now IntelAgent.

We saw an opportunity to automate a lot of what we were doing and in doing so increase our efficiencies internally. A couple of years back we started toying with the idea of turning that into an actual product to roll out; firstly to our existing customers at the agency and then beyond so that's what happened. We spun We Build Bots out of the agency last year and haven't really looked back since.

MO:  So, We Build Bots flagship product is IntelAgent. Do you want to explain a little bit about IntelAgent and what it can actually help companies achieve?

PS: I mean the headline is save money, make money and delight your customers. So, if you imagine IntelAgent as a box of tricks that sits in the contact center, its cloud-based and it accepts customer connections, whether that be via email, messenger bots, live chat, web chat or voice, any of those constitute as inputs. For example, you could type into messenger; ‘I need to change my address’, that input then goes into IntelAgent, we integrate with internal systems and then will spit out an outcome that says ‘that's all done for you’.

That can happen across Google Assistant, Amazon Alexa and Web Chat as well. So, ultimately we're improving customer services using AI and Analytics.

MO: Yeah, great. That's how it started off isn't it, it was a chatbot functionality where people could talk to the businesses they were trying to access but over time it obviously became a lot more developed and we're doing Integrations into systems for the users.

PS: Yeah, I think a big thing we learned, probably from the agency, was how to reuse that data. Everything that happens through IntelAgent gets profiled so consequently marketing departments, for example could log into the system and they could say ‘show me all the people who have spoken to us in the last two weeks who were talking about a city break and who want to travel outside of school term time’. Obviously that example is a travel company. IntelAgent will then collate a list of those relevant people and then you can broadcast messages out if you have a a relevant offer for those people.

MO: Yeah, I suppose on that note, the important thing to mention is GDPR because I bet people's ears pricked up when you said storing information. We are GDPR are compliant though aren’t we?

PS: Yeah, absolutely. We’ve had legal teams working on that for us. That's all documented and it actually can improve businesses compliancey because you can document the conversation and you can audit what's been said within messenger or web chat etc so you can explicitly request permissions for certain activities that you want to do off the back of the conversations and that is locked forever in time.

MO: Yeah, that's good. So that's a little bit of an overview of the product. Now let's come back a little bit to some of the stuff I mentioned around the Oracle program, I think it’s a few thousand companies that go for that, is that right?

PS: Yeah. I don't know the number really. It’s Global and there are other cohorts in Tel Aviv,  Sao Paulo, here in the UK, maybe Berlin. There are seven or eight cohorts around the world and it's a great program.

So yeah, wouldn't be surprised if thousands of companies try to get on it.

MO: They pick six companies from the UK and we were one of the companies that were picked this year. How are you finding that the accelerator program has helped us so far / do you think it has helped us and do you think that's going to help us going forward?

PS: Yes, I think it's definitely helped us. They’re very hands-on in terms of the shift over to their infrastructure so we'll be moving over to the Oracle Cloud. They've got some great new products like function as a service which we're running some of our AI stuff through - they've been very helpful with getting us onto that platform.

From a mentoring point of view, they've been very good. We've got to speak to some very high-level people in the company. It's really good to look at the journey and see the kind of the bits that we can take from that even though we're a way smaller startup, ultimately we’re a cloud software company so learning from one of the biggest and one of the best is fantastic.

Ironically, when we were writing the business plan about eighteen months ago, we looked at the three to five-year exit horizon and who might be interested in a company like ours if we got things right, Oracle were on that list. Strategically, it's potentially hugely significant because if we can nurture the relationship and build products and micro services that Oracle can use and feel are a good addition to their arsenal then the relationship could deep beyond where we are now.

MO: Yeah, that’s brilliant. For anyone who's listening and maybe doesn't understand the term microservices, it's a way of people being able to hit a service that we create on an individual transaction basis rather than it being built into a fully-fledged solution that they would use on a regular basis isn't it? So it's something that people can utilize across loads of different companies.

PS: Yeah, I guess it's a modular approach. We've built a platform that does lots of different things and those different things are called microservices. Swarm is one of them where you can create groups of similar looking customers. Agent Handoff is another one whereby we can alert agents if a chat is perhaps caught in a loop or needs a human agent because there's a deeper level of complexity than can be handled by something like a chatbot. Upsell is another one whereby we have a propensity to purchase algorithm, so if it becomes apparent that someone in a chat is potentially in the market for a service that our company offers then IntelAgent goes and grabs that product or service if you like and feeds it into the chat to say for example; ‘Hey, Marco, you might be interested in this based on what we've been talking about’.

MO: Yeah, that’s really  good. So, what would you say are the biggest misconceptions around chatbots? When people hear the word chatbot, there's an instant trigger in people's minds and what do you think they need to think about when someone says that word?

PS: Yeah, I think it depends who you're talking to, from the completely uninitiated place - like my mum for example, she knows a bit more about it now to be fair but I think the word bot immediately conjures up images of actual robots. So that's one side of things. I think the people who know chatbots probably feel that they’ve been overhyped and the hype needs pairing back and that's starting to happen. Then in the middle, there are people who use chatbots but they use them in a way that the chatbots can't handle, for example, just free typing and expected a chatbot to be able to handle a holy natural conversation. Instead of saying ‘I have a question about my bill’, they might  say ‘The post arrived this morning and I was a bit tired so I opened the bill and I didn't realise until later on today because I had to go to work all day and then drop the kids off to school that the bill was wrong”. Now, we would kind of pick that up but it's just an example of people feeling that they are chatting to a human as opposed to a bot.

I think that will pass and I think people will realise that if you've got a task to execute you can come onto a platform like ours or there are many others out there and you can execute that task really quickly. It's just a learning process of people understanding what a chatbot understands and while they do that I think chatbots get more intelligent anyway. There probably will be a point where people can come on and ramble.

MO: Yeah, I mean we can only go so far with it at the moment but who knows what will happen in the future. We are even at the point now where we’re isolating keywords, spotting the topic, analysing the sentiment and we're finding all of that information so that's really powerful, even compared to where it was a year ago.

PS: Yeah, absolutely. It's always moving forward.

MO: So what are the plans for our product  going forward then? What's on the horizon? Is there anything you want to  share with people around the direction we're going and where you see the company and IntelAgent developing in the future.

PS: Definitely, the product is always improving whether that be additional microservices or just improvements to the core engine so the natural language understanding and natural language processing - the AI. So all of that stuff is continually improving whether that be from a standalone machine learning point of view or assisted machine learning. We're always feeding the data that comes through the platform back to see what improvements can be made and then as I say, there's a number of ideas in the pipeline around additional microservices. Deepening the relationship with Oracle over the next four to five months is strategically quite key.

We’re also talking with some other partners at the moment who we might work with on a reseller basis. Voice has always been a big part of what we do and it's becoming a bigger part of what everyone does so we want to be right at the front of that movement to the point that we are currently developing our POC intelligent IVR System that will be showcasing next month I think. To help finish development of that product and to help continue on the path that we’re on with We Build Bots as a company, we're about to launch into our Series A fundraising round from next month and hope to have that closed in quarter one 2019.

MO: Exciting times - busy times.

PS: Yeah, this is very busy times.

MO: So in terms of the IVR, is there anything you can give details on around that or a bit more insight into what we mean by IVR.

PS: Yeah, so IVR stands for Interactive Voice Response and the listeners will know it from picking up the phone and an IVR system says ‘press one for this, press two for that’ or some of the the more modern ones will now say; ‘why are you calling?’ and you can say things like ‘Bill’ or ‘Returns’ or something like that. None of this is hugely natural though and none of it is hugely effective.

The reason we started building IntelAgent and using chatbots was because there's a huge channel shift happening from telephony into text based messaging. That could be WhatsApp, that could be Messenger, before that I guess it was Web Chat. That shift is happening and continues apace but still lengths ahead in terms of the channel that people use to contact companies - it's still telephone. That's still the most popular channel in terms of numbers but IVR Systems are also the most frustrating channel/platform that customers use. If you put those two together, people are still using telephone in their millions and they're all frustrated with the platform that they hit on the telephone and that screams out for disruption.

We're taking all of our learnings from IntelAgent and taking the core of that platform and moving in to disrupt the IVR market, which is about a six billion dollar market - it's a big chunk of business to be had if you can get it right. It's not going to be easy and it's not going to be fast but yeah, that's what we are looking at over the horizon. That's the next stage and that's one of the reasons we're going back out to raise from the market.

MO: It's exciting to see where it could possibly go in two years time and see where we are then compared to now.

PS: Yeah. If anyone searched Google Duplex, they’ll see a video of a really good demonstration that Google have done. We've kind of flipped that on its head. Google have done/are doing something with duplex whereby you can ask your Google Assistant to phone say a hair salon to make an appointment. In fact, the demo shows that; it's an assistant talking to a human but it's a very natural conversation.

For us, we think it makes more sense to flip that on its head and the more likely scenario and the more useful scenario unless you're trying to sell phones like Google are, is that a human will phone and get connected to a voice agent - a voice bot and they can say something along the lines of ‘I'm on my way to the airport and I forgot to check in, can you help me?’, through voice authentication and through IntelAgent Integrations, we will be able to say ‘Yeah, no problem Marco’ we’d run through a couple of questions like you do on any online check-in and then send the boarding passes directly to your handset. So that’s that. It would be very cool.

MO: Yeah, that would make life easier for everybody wouldn’t it. If we could get that up and running and in position that would be fantastic.

PS: It would, 100%.

MO: So, the majority of startups fail and unfortunately that's mainly due to lack of funding. Obviously, at We Build Bots, we manage to achieve 55% of the original target in 10 days which is unbelievable and then ended up over funded by 45%. Can you give the listeners a little bit of an insight into your experience and any advice you might give to startups going through their first round of funding.

PS: Yeah. Well I can talk about my experience. I'm no expert but I can certainly talk around what we did and how we went about it. We were looking for £500,000 which transpired, in our experience, to be a bit too punchy for some of the Angel Investors and the networks that we were talking to but not big enough for some of the Institutional Investors. The Angels typically wanted to push us down on valuation and wanted to put in around about £100,000/£200,000 between a consortium and consortiums are notoriously hard to manage because you've got to get everyone to agree. Then the Institutional/VC type money don't really, in our experience, open up conversations before two/three million which we just didn't really need at the time.

The other route was crowdfunded. I know some of the guys at Seedrs which was the platform we used and there's other ones too like Crowdcube but we went with Seedrs and it seemed to plug the gap really well. We managed to secure what they call a Cornerstone Investor, Seedrs typically like you to come on their platform with about 30% of the money raised already. We managed to do 50% through Development Bank of Wales, which is great and then we got off to a really good start on Seedrs. It started to tail off a little bit in the middle of the campaign - it's a 60-day campaign so around about 25/30 days it starts to tail off a bit so we just turbocharged the campaign. We went back out to our existing networks, did email marketing, put some Facebook ad spend behind it, did some PPC and just became a bit more aggressive in the way we did our outreach. Then we got a big chunk dropped in by an Angel and from there things just really accelerated.

We hit our target with about 20 days to go and we just let it run and we closed on about £720,000 which was about 40% overfunding which was good actually because it meant that we could accelerate some of our hiring plans within six weeks. Then we went from four of us to fourteen of us.

MO: Myself included.

PS: Yep,  yourself included. You’re probably the best one.

MO: Thanks very much, nice of you to say that.

PS: Yeah. I'm actually glad that we did over fund as it's extends the runway. It means that we can get moving a bit more quickly than planned and that's ringing true, we're ahead of target and touch wood so far so good.

MO: Yeah and like you said, I think it's going from strength to strength and we've got the next round of funding so if people want to follow the journey they can add us on Instagram, Facebook, Twitter and LinkedIn or visit our website at IntelAgent.ai.

PS: Yeah and if there are investors listening or Angels or anyone who wants to be involved in the next round then watch this space, we’ll be going back to market within the next four weeks.

MO: Fantastic. So, thanks for joining - one of the things we ask everyone is what tool or app couldn't you live without if you had to remove all your other applications?

PS: Probably WhatsApp. I’m in so many groups, whether it be work, social etc. I've joined some really interesting ones from voice groups to digital event groups. I go to Singapore next week and there's a group around that for what to do in Singapore. Also family groups of course.

MO: Of course, there’s always a family group.

PS: Yeah so probably Whatsapp. There are probably more important ones out there but that's the one I’d choose.

MO:  It's connecting you to other people isn’t it.

PS: Yeah, I think so. It's the one that springs to mind.

MO: Well, thank you again for joining us! We’ll catch you soon.