everythingpossible - Fotolia
Lucidworks claims to be on the frontline of search, by marrying data with the expertise of people in a growing business that helps ecommerce companies and financial firms alike to deliver more relevant search results and customer experiences.
Last year, the San Francisco-based company secured a new round of funding, fuelling its international expansion plans while continuing its work in areas such as deep learning and question-and-answer systems.
In an exclusive interview with Computer Weekly in Singapore, Lucidworks’ CEO Will Hayes discusses his business priorities in Asia, the current limitations of search technology, and how it is helping enterprises break down data silos.
Lucidworks had raised $50m in Series E funding in May 2018. Did any of that go into supporting your international expansion plans?
Hayes: At the time, the company was growing at 120%, so there was a lot of excitement and energy around the potential of the market. We felt it was time for us to secure additional capital and focus on tapping that potential to provide more leverage for the business.
We were reaching a point where the demand was outpacing our ability to execute. Within nine months, we’ve doubled the team to about 250 employees. International expansion was part of the plan, but at the time, we still were building out our go-to-market capabilities around the US and parts of Europe. Now, we’re starting to look towards the Asia-Pacific region.
You will start to see us building the team around Robert Lau, our chief operating officer for global emerging markets, and you are going to see us make a real push into Asia with a number of offices and some strategic partnerships along the way.
Are you able to provide examples of Asian markets that you’re planning to invest in? And how do you see those markets from a strategic perspective?
Hayes: If I just take a step back and look at global trends, we see the Asian market driving a lot of innovation. And it’s an interesting market because it isn’t typically defined as a first mover.
Historically, a company would take the same playbook from North America, with repeatable and referenceable examples, to Asia and start telling its story. But in areas like digital commerce and mobile engagement, Asia has surpassed and adapted so much quicker than North America.
With our excitement around the innovation in Asia, we’re bringing the combination of what we’re doing around machine learning and AI to the region. This isn’t a science experiment or pie in the sky. This is about ensuring that a customer who visits a website engages with a company and maximises the value of that visit.
Specifically to your question, when we look at the various markets, there are obvious ones like Japan, Korea, Greater China and Singapore. But we don’t go into those markets alone – we rely heavily on our partner communities as well.
Who are your typical customers? Can you provide examples of use cases and the most common problems that you’re solving for them?
Hayes: For the last couple of years, we’ve been developing our platform called Fusion that takes the concept of search and marrying that with machine learning modules that can help with very specific things around data enrichment and information retrieval.
An example of this would be document classification and clustering. We used to do that through ontologies, but we can now do that programmatically. On the retrieval side, we can personalise search results based on a user’s behaviour. We can dynamically rank data for you. Fusion is a $40m product which has been incredibly successful for us.
A key area for us is the digital workplace, where we intelligently find relationships between entities within content and recognise that someone is an expert in a particular domain. PricewaterhouseCoopers is using our platform to connect 40,000 consultants around the world, while Morgan Stanley is using the smarter data to empower financial advisers so they can serve their customers better.
Another area would be digital commerce. We work with three out of four of the major home improvement retailers in the US. These companies are looking to compete by using better insights around data to service you as you’re discovering, shopping and transacting.
Will Hayes, Lucidworks
So as you’re checking out, they can send you promotions and recommend additional products. On the retention side, they are able to understand behaviour and preferences, and support you better as a customer, like sending you a particular promotion or knowing that there’s a defect with the product you have.
Our system is helping these companies compete in a world where they’re under attack in terms of price and convenience. If they tried to beat Amazon or Alibaba on convenience and do same-day shipping, they can forget about it.
But experience is something these companies can hone. They know their domains through their merchandising and product experts, who understand that customers who like a particular kind of device may have a preference for something else. With that knowledge, they can build workflows that are specific to a customer.
Open access to data
For this to work, you need open access to data. In many cases, organisations don’t even know what data they have because of departmental silos. How do you break down those barriers when you go into a project with a customer?
Hayes: The first thing we do – and this is a common challenge for us – is to start with the customers and say, “Forget about data – why are you talking about data? What are we trying to do?” If they want to service customers better, we have views on how they can do that.
First, we need to understand the information in their customer relationship management (CRM), as well as product and issue information, including bug tracking and service tickets. We also need to understand things like catalogue availability and inventory to define a data model. We like to come in with very precise answers, and I think that creates a level of comfort for the customer to be willing to collaborate with us to achieve their goals.
Now, the customer might say they can’t give us some of the information or that it’s too complicated because they have to work with another team in another country. But through our experience, we can know we make things better if the customer brings in an additional dimension.
And we show them the proof, the metrics, and the things we’ve seen in the industry that we know will work for them. This is one of the easiest technologies I’ve worked on because the focus is purely on business outcomes.
Understanding AI decisions
You talked briefly about AI earlier. How are you helping customers to better understand the decisions made by AI through your platform?
Hayes: We could look at this in many ways. We could go into the social and responsibility aspects, but we’re not deciding if somebody lives or dies, right? We’re not even trying to go power a vehicle or something.
Well, it could be making recommendations that might not be good for you.
Hayes: Precisely, so there’s that. We believe in supervised learning, and frankly, you have to in order to be effective.
One of the jokes about Amazon is that Amazon has no taste. You look for a lawnmower, and all of a sudden, you get recommendations on running shoes. Now, those might be the right running shoes and you might buy them anyway, but that’s not an experience, right? It’s just a bunch of recommendations being thrown at you.
I strongly believe you need to have that supervised sort of hand-off because without it, you’re missing out on people – a key asset within your organisation. On the responsibility part, it is my personal philosophy that more and more transparency with data is going to become more important than data protection.
Who would you consider to be your biggest competitors? I was struggling to find a company that might be your competitor.
Hayes: Yeah, no one can figure it out. To be honest with you, in terms of what our platform provides, which is really about connected data experiences, we feel strongly that we’re creating a category that has not been defined yet, and we can prove that through customers, investments and validation.
If I look at individual use cases, then that is where competition comes up. There are a number of e-commerce search vendors out there – Oracle, SAP Hybris and so on. But from an innovation standpoint, we don’t really see a problem. We’re seeing fewer and fewer cases of us losing a deal because a potential customer has to go with Oracle or IBM.
There are some start-ups in this space as well, but they tend to be a little more down-market. For us, we want to serve the entire user journey, which is going to be enhanced by data from stores, channels and multiple countries in which they operate. We can provide a ton more value with all of that data.
That said, we’ve had smaller customers that come to us with over $1m for an annual licence. I’ve never seen that in my career, but I think they’re buying into where they need to go as a business and what we could provide, and I think that’s really unique.
What are some of the R&D areas that you’re looking at?
Hayes: We’ve been leaders around information retrieval, and where AI and machine learning have played a role in areas like relevancy and ranking. Obviously, TensorFlow and deep learning are really interesting and exciting to us – we’re going to announce a partnership around that pretty soon.
Read more about enterprise search and content management
- The enterprise search market is evolving and offering new options, such as open source and AI. Discover the benefits of these platforms and how companies are using them.
- You can search text, but it’s harder to catalogue images and video. New tools are automating those processes at the enterprise level.
- Unsung and unheralded, semantic technology is a key component in artificial intelligence and other big data applications. Yet, like AI, it still faces hurdles to going mainstream.
- Web content management platforms, customer data platforms and matchmaking can all help companies create and implement a personalized customer journey savvy consumers demand.
Another area is question and answering – we have some prototypes and beta programmes today with some customers. The challenge with those systems is that they have been very domain-specific, so we’re starting with insurance and financial services. That paradigm will continue and we’ll be able to expand in other areas. We also think that deep learning is going give us more shared learnings between these domains, but that’s not been proven yet.
What you just mentioned is a very crowded space, isn’t it? The chatbot guys are all doing that, aren't they?
Hayes: You know, they are but they aren’t. What we see is that there are a lot of parlour tricks out there. You could put some rules together and call it a chatbot.
We’re not necessarily trying to embed ourselves on a website, but what we’re finding is it’s more about query classification and understanding – which is right there in our world.
We’re sitting on top of the data and processing it in a very intelligent way. We understand ranking of data and behaviour around data. We understand the context of what you’re asking, and map those verbs and nouns back to our ontology to find the answer.
There are some chatbot providers that have reached out to us because they want a smarter data store to retrieve from. I don’t think any of them have really cracked that, and, again, I think we’re at the forefront given what we’ve already accomplished within enterprises.