Using AI to provide a more powerful search experience can be difficult.
Lucidworks, a provider of AI-powered search apps, is trying to simplify the process and provide businesses with a connected experience with a new SaaS platform called Springboard.
Thursday’s release includes Springboard’s first generally available app, Connected Search.
Connected Search provides businesses with a search and information engine featuring push-button AI and guided, optimized workflows.
Companies that use Connected Search use Springboard’s artificial intelligence and machine learning capabilities. Lucidworks said push-button AI provides instant intelligence and improves relevance by applying user information and cues.
Lucidworks plans to release additional apps for the Springboard platform throughout the year, including Connected Service in Q3 and Connected Commerce in Q4.
The Connected Search app is listed to start at $600 per million requests and 100,000 documents per month for early access customers.
The platform’s goal is to enable businesses to understand the intent of their customers, CEO Will Hayes said.
In this Q&A session, Hayes explained what these connected experiences are and how Lucidworks’ new search platform differs from competitive products.
How was this research platform born?
Will Hayes: For the past eight years we’ve been working with what we call the fusion server. It is an AI-powered search engine with a primary focus on personalization and discovery. We take all user signals while you shop, while you browse, add things to your cart, contact user support, we build a comprehensive catalog. We run machine learning to use these signals to better target personalization and categorize things for you. This is used by some of the biggest retailers, some of the biggest grocery, home improvement, auto renovation, clothing and exercise equipment companies.
We’ve received many requests to take these features – this AI-powered search, personalization, and recommendations – and apply them to additional use cases like customer service and e-commerce.
Will HayesCEO, Lucid Works
To make this faster and more impactful for our customers, we are launching a SaaS platform. It is an application-specific multi-tenant SaaS platform.
What problem are you addressing with this research platform?
Hayes: The biggest differentiation is how we leverage what we call these user signals.
Unlike many competitors who focus on a particular type of search use case, we believe that by bringing all of these signals together, by running machine learning on all of these signals, we can start informing multiple channels.
One of the biggest differentiators and benefits companies get from connected search is that they start collecting all that user interaction data; this is what we call first-party data.
[Connected Search] is the site search application. This will allow customers to respond to your browsing experience and allow them to change the results.
What makes this search platform different from your competitors?
Hayes: There are three types of key elements. One of them is how we applied AI within the technology. We have several different machine learning models that we use out of the box, for visualization and for recommendations. We have some who are trained on very domain-specific types of things. If you’re in oil and gas or financial services, our AI can help better enrich data, understand data, answer user questions.
So when people come in and ask a question, you know, being able to surface and answer those questions, so we rely on something called semantic vectors.
Then if you look within our platform itself, we kind of have two key paradigms that we rely on to deploy these use cases.
The first is our data services. Again, I talked a bit about being specifically trained for specific areas. We go into oil and gas or apparel, or home improvement. These all have their own sets of languages in the glossaries they use around which we can enrich and understand the data. Then, with our workflows, we’re allowed to take all that magic, if you will, and put it into a process or flow that, for example, can help support a customer service agent.
Here is a simple workflow example. I’m a help desk agent; I receive a phone call. I type [the customer’s name]. This workflow will know how to extract all products and services related to [the customer], retrieve all open issues related to these products and services, find a resolution if we have a common issue that occurs, and display all this information at the time to provide this workflow.
From customer agent [perspective], I just go to the screen, and I say, ‘Oh, okay. I see the things you subscribed to. I see some issues you might be having, I see where you’ve been on the website, let me try to serve you better.
By bringing these things together – semantic vectors and AI, our data models that we’ve built around various domains and these workflows – we have a unique way to go to market and deploy different use cases , using these same abilities.
How quickly will users be able to implement this enterprise search platform in their workflows?
Hayes: There are two things about speed. One of them is sub-second indexing.
So you give us a URL, we go in, we start calling incredibly fast. We also update quickly.
In most cases, with most solutions, there may be a delay of up to 12-24 hours before these updates appear in the index. Imagine if you’re like a grocery store supplier, and inventory is constantly changing and you need to keep those indexes up to date. We can do it in real time, and we can do it on a massive scale.
Editor’s Note: This interview has been edited for clarity and brevity.