SearchBlox

The Company

SearchBlox was founded in 2003 and is headquartered in Glen Allen, Virginia. The company is bootstrapped and led by co-founders and brothers Timo Selvaraj and Robert Selvaraj. SearchBlox has deep expertise in the enterprise search world, with over 600 customers of their core search engine product and managed services running on AWS. However, the company now has two core products, Enterprise Search (aka SearchBlox 4S) and SearchAI, a newer suite of products launched in 2019. This report focuses on SearchAI.

The Technology

SearchAI is, of course, an AI-enabled search technology. Or rather, it is four technologies that can work together:

• SmartSuggest
• Answers
• Personalization
• Chatbot

Before looking at each in turn, it’s important to note that all four modules run on a common core platform. At the base is the OpenSearch platform, which means that SearchAI also runs on and can easily be managed within the AWS cloud. On top of this is the usual assortment of search mechanisms such as parsing and ingestion tools, connectors, and crawlers. This functionality is further enhanced by SearchBlox’s own trademarked PreText Natural Language Processing (NLP) technology. The PreText NLP service can translate audio to text, generate titles and descriptions, provide sentiment analysis, and undertake automatic tagging/classification. In addition, there are over 300 built-in connectors.

In other words, underneath SearchAI is SearchBlox’s enterprise search engine core product. This product, Enterprise Search, can be bought standalone and is popular with smaller to mid-sized organizations, initially replacing the now-defunct Google Search Appliance. SearchAI is a set of add-on modules that run on the same infrastructure foundations. This is not a bad thing – far from it, as the core SearchBlox engine is solid and well tested. Instead, the company is developing standalone and specific (rather than general) intelligent applications.

To date, there are four SearchAI modules:

• SmartSuggest personalizes suggestions as you type in your search, through an understanding of your working context and content. It does this by utilizing a number of intelligent algorithms to replace traditional auto-complete.
• Answers builds further on the concept of SmartSuggest, providing answers to employee or customer questions by leveraging the same underlying NLP mechanisms.
• Personalization provides both recommendations and search results to employee and customer inquiries based on predicted personas.
• Chatbot is the module for voice or text.

These SearchAI modules run on a second common platform that uses machine learning to understand content popularity and add a degree of personalization to gradually filter and narrow down result options. These filtered results then feed through a more advanced AI algorithm that continually learns and, in theory, improves the quality of the results it provides over time (see Figure 1).

There is a theme here: to make search more of a push function (pushing the correct data based on prompts) rather than a pull function (typing in a search string and getting the end user to choose from many options). This trend cuts across the world of search, but as of today, it is typically very expensive and complex to do. SearchBlox is relatively inexpensive compared to other enterprise search or insight engine vendors and to a large degree comes out of the box.