Paul could you tell us more about the WeBuildBots story?

It started with a social and mobile agency I founded in 2010 called Coup Media, which had incubated WeBuildBots for the last 18 months.

Apart from the development of social campaigns and mobile apps, we also managed a lot of the social media customer service channels for companies like GoCompare, Principality and The AA. In doing so we started to identify the repetitive nature of customer queries; about 40 percent of the questions we were answering were about the same stuff.

So we started building some rudimentary FAQ social applications, redirecting people from things like Twitter and Facebook feeds into a searchable knowledge base apps. That worked well, and we saw some efficiencies there.

We started moving more and more into automation, and so when Facebook opened their messenger SDK in April 2016, the stars aligned to a certain extent; we had the social media and mobile experience, we had the customer service experience, we had the technical know-how and we had a list of existing clients. So we started going back to our clients to ask if they’d like to see IntelAgent, the customer service chatbot platform we’d developed. The take up was so encouraging that we spun the product out as a separate business: We Build Bots!

Our flagship product is IntelAgent, an enterprise level customer service platform which uses chatbots as the customer interface. An underlying AI suite automates tasks like accessing customer account information to facilitate self-service, identifying sales opportunities, engaging the most appropriate human agents if and when intervention is needed, and constantly monitoring and feeding back ‘health of the contact centre’ data to those that need it.

An example of an improved customer experience through IntelAgent is this:

Imagine you want to request a bill from your gas provider. You could phone them, wait on hold for 6 minutes (the average time for phone channels), and if you’re lucky they may be able to email it to you within 24 hours More likely, they’ll post it to you which can take 5-7 days. That’s a pain. Some companies do have online portals but they see incredibly low traffic rates because users need to find the portal via the company website, remember their log in, find the section in the portal which houses bills and then download them. That’s not a great experience either!

Through something like IntelAgent, you open your provider’s Facebook Messenger Chatbot, tell it you’re looking for a bill, pop in your email address and the date of the bill you need, and it sends the bill directly though Messenger. For download. The whole process takes less than 90 seconds.

We’re not just making the customer experience really slick here, we’re actually saving companies millions of pounds each year because their agents are no longer spending their time on repetitive, non-value-add tasks. Consider that a human agent costs about £24 per hour (when you incorporate things like software, lighting, desk space and salary), and the average contact centre is 125 people strong… it’s easy to see where the costs quickly add up. We’re removing those costs and improving the services companies offer their customers.

So in terms of how transferable IntelAgent is, can you guys utilise the data behind the scenes and provide almost a consultancy piece to companies on things like root cause analysis for example?

Yes. The upfront consultancy element of an IntelAgent integration sees us extract from the company any customer contact data it may already have; transcribed phone calls, webchat logs, social media complaints, email, knowledge bases etc. We take all of that data, and pour it into our AI and data engine and we start to find the pain points, trends, common issues, most requested tasks and more detail.

That gets us to a point of a well-trained chat bot for that company, to build on their data. Post launch we continue to collect more and more conversational data, which now includes not just free text but also what buttons and menus are used, how quickly people are responding, the sentiment of users, what words come up frequently, the gender, location, and how all those data points map against each other to give a really clear view of customer interaction with a view to making predictions around the type of support any given customer is likely to need and preempting that need.

So to give an example, one of our customers recently, found the word ‘still’ kept cropping up, it was people staying ‘I’m STILL waiting for…’ this particularly client is a claims processor, so off the back of that we recommended they should push ‘claim updates’ via the chatbot so people aren’t left waiting. We’ve seen an 80% reduction in those types of complaint over the last 3 months and the client has experienced a 10% drop in overall call volume.

So we do all of that as a continual improvement process, and that’s one of the things that sets us apart; we collect the data and resolve issues or find opportunities for efficiencies using AI, human interrogation and Machine Learning.

We did an event recently around the Pros and Cons of Artificial Intelligence, so I’d be interested to know do you get much resistance to Chatbots maybe just as a pure concept?

Interestingly not very often. Certainly very little from an ethics perspective. The closest we have come to that was when we were nearing the completion of a deal, and the client engaged their HR and internal comms team. Their opinion was that as soon as you mention chatbots in a call centre, it sends a certain message internally and people instantly assume their jobs are under threat. The truth is humans and AI work better together than either of them can alone.

The other objections are more around ‘do we tell people we are using chatbots or do we pretend to users its a human?’ (to which we always say: ‘No, please don’t pretend its a human’).

Other questions have been around the use and adoption messaging platforms, with the biggest concern being ‘What if my clients aren’t on Facebook Messenger’. The answer is A) They probably are or B) your customers didn’t used to be on Twitter or the web or on mobile apps but they are now, and the more you can encourage them to use Facebook Messenger the bigger your savings are going to be and the happier your customers are going to be.

The growth of Messenger is huge; from around 200million users in 2014 to 1.3billion now. Around 2 billion messages sent between individuals and companies every month, and we often relay these stats and try to impress on clients what a major channel Facebook Messenger is. Messaging has become a bonafide way of communicating with companies, and in 2015 it became a bigger online activity than social media browsing… which is huge!

There’s a messaging revolution happening and it’s accelerating as the younger generation become the generation wanting to deal with telco’s, utility providers, banks, retailers etc. Its a huge trend that’s going one way so even if your current customers aren’t using messages services (unlikely), your next wave of customers will be, so get ahead of the curve now!

So do you think that the acceptance of AI is growing? For example, if we had asked that same question 3 years ago how different do you think it would have been?

I don’t think there would have been a fear about AI so much, but I think there would have been a much bigger education piece about what AI is. I think people would have jumped straight to the movies and robots! It wasn’t such a common (overuse, overhyped) term back then so mentioning it would have created more confusion than it does now.

Its flipped the other way over the past 12 months whereby now its kind of being overused.

Everyone throws around that AI term now when there are not necessarily doing AI. To a certain extent it’s relative; compare us to a standard chatbot agency and yes we’re totally doing AI, but compare us to someone like Hanson Robotics, and maybe they’d argue what we’re doing isn’t AI.

Why do you think that is?

I think it’s akin to when everyone started adding .com to the end of their logos. Being a .com startup was super cool during the original explosion of tech companies and I think AI is maybe going through a similar cool phase. In a way at least it gets the term out there so when we talk to people about AI, they don’t then jump straight to robots!

How do thing this conversation would be different if we did this conversation in a years time, how do you see this evolving?

Chatbot adoption will have started becoming mainstream for sure.

I think voice will be massive by then too. We’ll definitely be doing more in voice recognition helping customers talk to their handsets or home assistants.

The other thing is… we don’t know – and thats the exciting bit. We’re collecting some very valuable data and we can do some clever stuff with that data already. In a years time we will have a lot more data so will have uncovered new, sector specific trends we didn’t see coming, and being able to roll our findings out to our clients, helping them drive new efficiencies and new innovations through AI… that’s really exciting!

What tech sits under the hood of IntelAgent?

Its all in the cloud, sat on AWS, and we have a mixture of propitiatory technology plug in stuff.

Our proprietary functionalities include ‘Swarm’ ‘Upsell’ ‘Handoff’ ‘Streak’ and ‘Look and Learn’. These modules cover stuff like customer profiling, propensity to purchase identification, call routing, customer loyalty and triggered machine learning. Then we also integrate with IBM Watson, and a sentiment tool we built for an earlier project.

How much has the major players opensourcing all of their data contributed to the data hype late?

Were not pulling on open source data just yet, but we definitely can do. I was involved in another project that utilised social data, and it correlated cinema box offices takings to social data. And we started to pull in bits of data around traffic, and weather, and so if its a hot weekend how does that affect the box offices takings? So I think its bound to have had a massive affect, as well as things like Kaggle and InnvoateUK running competitions based on open data. Open banking data too will see a surge in Fintech businesses I’m sure.

The other thing is that companies are more aware of running hackathons and the like, and getting people to play with their data. That can often lead to innovation. Indeed bigger brands – and increasingly banks – are investing in in-house and external incubator programmes – Unilever Foundry is one of the major examples.

Where has your interest in analytics come from?

I’ve always had a real interest in ‘what’s next’ tech wise in the business realm. When I set up Coup Media 8 years ago, social and mobile was relatively new (certainly in terms of business usage). I was lucky to have been exposed to that trend quite early through my work in the US with companies like Time Warner, Google and Bank of America.

When I came home I set up Coup to deliver social/ mobile strategies, campaigns and applications. The way we proved these channels weren’t just fads or a ‘nice to haves’ was through deep analysis of the results so we could relay findings to decision makers.

Over the years we have moved further and further into the analysis of social /mobile data and the development of software that interacts with that data.

That led to the establishment of a second company called Hello Soda which looked initially at social data in credit scoring. Soda has gone on to employ around 35 people and recently raised £5million investment.

Now We Build Bots is a culmination of my learnings from the past 8 years – social, mobile, customer services, data, analytics, and AI.

How do you find being a tech start-up based in Cardiff?

Tramshed is absolutely fantastic. We’ve been here for about a year now, and can see it’s really becoming a hub. We regularly have events going on here, investors coming in, we’ve had BBC here and Pitch at Palace. Its great to have a place in Cardiff that attracts people into the city to see and hear about what is going on.

I’ve become more aware over the past year of how many promising tech start ups there are in and around South Wales. I still think there is a stigma attached to Wales, (which is finally beginning to erode) whereby people maybe look at us and they don’t have the believe in what can be achieved here.

We are also based in Engine Shed, Bristol and I do think they’re some way ahead ahead of Wales. Their success stories are dwarfing ours at the moment (for example Graphcore have raised nearly £100million in funding) so we still have a little catching up to do, but we are certainly on the right track!

Its definitely a satisfying feeling being part of that journey with some of the other companies here, and I really hope it continues to thrive and live up to some of the hype.