Lucidworks this week made generally available an orchestration engine for artificial intelligence (AI) models that is integrated with a search platform widely used in e-commerce applications.
The goal is to make it simpler for organizations to integrate multiple AI applications and agents rather than having to acquire, deploy and manage a separate platform, says Keri Rich, vice president of product for Lucidworks.
In contrast, the orchestration engine provided by Lucidworks makes it easier to connect and normalize data that AI applications require, and then apply a set of guardrails to ensure AI models can only access specific sets of data and behave as expected.
In addition, Lucidworks plans to add low-code/no code tools that will make it simpler for teams to build AI workflows without necessarily requiring to have deep AI engineering expertise, says Rich.
Lucidworks has been making a case for a search engine platform that, for example, is optimized for searching product information on an e-commerce site. The Lucidworks AI platform adds a generative AI framework that is being used to drive significant improvements in search relevance, with the company claiming customers reporting 2.5x more successful AI deployments – a 12% increase in relevant result rankings, and a 90% reduction in zero-results queries.
The orchestration engine will make it easier for organizations to combine different types of AI models to optimize e-commerce experiences, says Rich. In fact, one of the most common issues organizations are encountering when seeking to operationalize AI is that they wind up using the wrong type of AI model for a particular use case. The orchestration engine will make it easier to experiment with multiple large language models (LLMs) and machine learning (ML) models to ensure the one that drives the most optimal result is employed, adds Rich. “Organizations need to more carefully evaluate the return on investment in the AI models they decide to use,” she says. “In a lot of cases, ML is still the better choice.”
Most of the challenges organizations face in the context can be traced back to a data quality and management issue, Rich notes. A recent report published by Lucidworks found that only one in four organizations have successfully launched an AI initiative in the past year, she says.
There is little doubt that more organizations will be successfully operationalizing AI in the coming year, but there are significant hurdles that need to be overcome before that. Ultimately, organizations will need to decide to what degree they want to build AI capabilities as opposed to leveraging platforms provided by other vendors that in many instances they already employ.
Regardless of approach, the race to provide these capabilities is on. Less clear is to what extent those capabilities are becoming table stakes needed to remain competitive versus a sustainable competitive advantage that can be maintained for an extended period of time.
