Lucidworks – Machine Learning for Search Engines
Table of contents
About Lucidworks
About four years ago, San Francisco-based Lucidworks was on the ropes. It was losing business for its open source enterprise search platform on Apache Solr to competitors like Elasticsearch. Negative press followed news of layoffs, with one employee at the time telling VentureBeat he’d be surprised if Lucidworks lasted 18 more months.
That was September 2014. Will Hayes was only a few months into the CEO role after serving a year as the company’s chief product officer. Before then, he had held several positions with Splunk (NASDAQ:SPLK), which offers slick software for searching, monitoring, analyzing and visualizing machine-generated big data. But Hayes had never occupied a C-level office before.
It turns out Lucidworks needed a product guy to help sell its product.
“We took a company that was about to go under, and we’re closing this year above $40 million in revenue,” Hayes tells Nanalyze. “We’re growing 130 percent. We’re signing up a big chunk of the Fortune 500. We have more than 300 customers on this platform.”
Fusing AI in a Search Engine
The platform is Fusion, built on Apache Solr open-source software but heavily reinforced with artificial intelligence. “We took the concept of the search engine, we invested heavily in machine learning classification … and we released a platform that is weaponized AI,” he says. “We have capabilities around serving data in a very contextualized or personalized manner that is very unique in this space. In fact, no one does it quite the same way we do.”
Anyone who uses Google or Netflix is already familiar with the basics of AI-powered search engines. Fusion returns highly relevant results first by inferring relationships within data. Take the example of someone thinking about traveling to Honolulu. A travel search engine backed by Fusion might also recommend Tahiti or the Maldives, since both locations share similar characteristics and experiences. But Fusion goes a step further, looking at the ways users interact with data and capturing that behavior. Perhaps the user searching for a Hawaiian vacation is also into food and wine. That might return a suggestion of visiting the south of France, where one can find beaches, booze, and bœuf bourguignon. “That’s behind the scenes of what we’re doing on behalf of your data and your applications,” Hayes says.
Build It and They Will Come
The proof is in the customer list: Clients range from Reddit and Red Hat to Wells Fargo and American Express to Verizon and Vodafone.
For example, take the case of Red Hat (NYSE:RHT), which provides open-source software products. Red Hat’s customer support portal was originally built on Google search, but a rapidly growing business was driving licensing costs sky-high while click-throughs were rarely more than 40 percent. Customers weren’t finding what they needed. Enter Fusion, which ended up saving Red Hat 91 percent in total licensing costs while boosting daily click-through rates on the customer portal by 212 percent. Its customers were also able to self-resolve problems more frequently, resulting in 50,000 fewer support tickets.
Last year, in another case study, Lucidworks partnered with popular social media news and discussion website Reddit, which has about 300 million users generating 70 million searches every day. That works out to analyzing billions of records and applying machine learning to understand the data and to deliver highly relevant results across more than a million different community groups.
Drawing Big Insights
Lucidworks serves a number of industries, including financial services, media, manufacturing, government, pharma, and law enforcement. Hayes says use of its search engine is split about 40-40 between customer experience (consumer) and business productivity (enterprise). The other 20 percent of business comes from the insights that Fusion can help generate from the huge amounts of data ingested, what Hayes refers to as “operational intelligence”. In other words, the AI is not just connecting the dots but it’s drawing a picture, using search-driven analytics and artificial intelligence to make sense of big data. In the case of a customer like Verizon, that might mean identifying service performance issues that affect its customers based on billions of data points.
“That’s not something you just throw into a database and put into a tableau,” Hayes explains. “We’re aggregating lots of multi-structured information. We’re having to sift and rank that information to determine what the most relevant pieces are.”
To ensure that aggregated data is relevant to the end-user, Lucidworks made its first acquisition last year. It bought Twigkit, a software startup out of Iceland that specializes in user experiences for enterprise-grade search and big data applications, for an undisclosed amount.
Raising Money for the Future
Lucidworks has technically raised about $59 million since it was founded about a decade ago in 2007 and last took in $6 million debt financing in 2016. However, Hayes says he hit the reset button when he took the reins in 2014, so the official war chest stands at about $23 million. The shuffle makes Lucidworks a Series D company playing in a Series B ballpark, as it prepares to raise a new round of money. Its $21 million Series D raise in 2015 included Allegis Capital, Shasta Ventures and Granite Ventures.
Hayes didn’t provide too many details on the new round but did say Lucidworks would raise at least $20 million in what is shaping out to be an oversubscribed round. While the company is looking to grow, Hayes says Lucidworks wants to maintain “practical growth”.
“We’re a unique value prop for [Silicon] Valley, and that’s reflected in the interest,” he says. “We look like a Series B company on paper, so we’re undercapitalized, undervalued.”
The 150-person company is also not losing money, and it expects to break-even as its fiscal year comes to a close this month, so raising money when you’re not desperate seems to attract more investors.
“I’ve been on the other side, where it’s not so much of a party,” Hayes says.
We’ll continue to follow Lucidworks to see if the good times keep on rolling.
Update 08/12/2019: LucidWorks has raised $100 million in funding to continue to expand in AI-powered search-as-a-service for organizations. This brings the company’s total funding to $209 million to date.
Share
