When it comes to generative AI, enterprises need to think big. Shaving a few seconds off the time needed to draft an email is helpful, but the journey to real value begins when you apply AI at the enterprise level. A new partnership between Accenture and Elastic combines technical expertise and strategic excellence, enabling businesses to build the data foundations for a successful AI future.

Optimizing search relevance with retrieval and reranking

Mayzak adds, “Deploying a vector database and transforming enterprise data into embeddings is only the first step in making RAG and LLM workflows effective. The real challenge lies in optimizing search relevance and ensuring that AI retrieves the most contextually appropriate and high-value information.”

To enhance retrieval quality, Elastic uses multistage retrieval, where an initial recall step using vector search or a combination of keyword and vector-based techniques, a hybrid approach, is followed by reranking models that evaluate the retrieved documents for accuracy, contextual fit, and informativeness.

“Elastic puts heavy emphasis on fine-tuned transformer models to filter out noise, ensuring that the AI system prioritizes the most useful, trustworthy responses,” says Mayzak.

Tools such as Learning to Rank also support result accuracy, whether at the individual or cohort level, giving organizations flexibility when targeting different audiences. As the volume of data increases, the system learns which features have the greatest impact on relevance, allowing them to be prioritized in the model.

Accenture takes an equally diligent approach to search relevance. Rodriguez says, “We spend a lot of our time evaluating RAG and generative AI applications. To achieve 90%–95% levels of accuracy, you need a holistic process that shines light into every corner of the process.”

A good example is Accenture’s AI-powered search “operating room” process, which brings together experts from various domains (data ingestion, query construction, prompting, business) to diagnose and resolve accuracy issues using automated and insight-driven methods.

Rodriguez draws a parallel with a neurosurgeon operating on a patient. “Experts act like surgeons, poking and prodding the application, while other specialists observe and analyze.” This approach enables the team to pinpoint and address obstacles to search accuracy, which often relate to data quality, context, or the way queries are formulated. Automated methods can then be implemented to monitor the performance of the application over time.

Elastic’s developer experience is also fundamental to the partnership. “Elastic prioritizes how developers move from initial setup to production deployment. We strive to provide everything they need to achieve results quickly,” says Mayzak. This includes tools like Elasticsearch AI Playground that streamlines the process of building prototypes and launching production applications.

Many organizations are already reaping the benefits. Reed, the UK’s largest recruiter, is using Elastic vector search technology to save employers 20% of the cost per hire. Korea’s leading IT services company, LG CNS, has deployed Elastic generative AI, boosting search relevance by 95% and accelerating retrieval by 50% as a result.

“Real industry reinvention demands deep intellectual investment, and that's precisely what the Accenture-Elasticsearch partnership delivers,” says Mayzak. “We’re combining data-led technology with deep industry knowledge to get generative AI projects into production fast.”

Rodriguez agrees with the need to deliver measurable business value. By combining Elastic’s AI-native search capabilities with Accenture’s industry expertise, businesses can move beyond the hype and into an AI-powered future that’s both transformative and profitable.

Learn more about generative AI on Elastic’s Search AI Platform, or start a free 14-day trial.

Source:

1. Seagate, “Seagate’s ‘Rethink Data’ Report Reveals That 68% Of Data Available To Businesses Goes Unleveraged,” 2020.

The release and timing of any features or functionality described in this post remain at Elastic's sole discretion. Any features or functionality not currently available may not be delivered on time or at all.

In this blog post, we may have used or referred to third party generative AI tools, which are owned and operated by their respective owners. Elastic does not have any control over the third party tools and we have no responsibility or liability for their content, operation or use, nor for any loss or damage that may arise from your use of such tools. Please exercise caution when using AI tools with personal, sensitive or confidential information. Any data you submit may be used for AI training or other purposes. There is no guarantee that information you provide will be kept secure or confidential. You should familiarize yourself with the privacy practices and terms of use of any generative AI tools prior to use.

Elastic, Elasticsearch, and associated marks are trademarks, logos or registered trademarks of Elasticsearch N.V. in the United States and other countries. All other company and product names are trademarks, logos or registered trademarks of their respective owners.