Event Report: The Key to Successfully Using Generative AI Lies in Enriching the Database

In this webinar, we focused on enriching databases to enable the effective use of generative AI in retail, wholesale, and manufacturing, and introduced specific methods and examples of initiatives.

Importance of Improving Data Quality in Generative AI Use

At the opening, we explained how important improving data quality is in promoting the use of generative AI.
To maximize the benefits of generative AI, the key is how much it can be applied to existing operations. According to Givery, which has provided support for implementing and using generative AI to more than 450 companies, two measures are effective for applying generative AI to existing operations: 1) linking it with company data, and 2) creating dedicated prompts for each job function.
At this event, focusing on 1) linking with proprietary company data, Lazuli and Givery held a session on the “methods for improving data quality,” which are essential for using generative AI without causing hallucinations.

Lazuli's Proposed Method for Enriching Databases

Lazuli executive officer Kitashoji took the stage and introduced specific methods for how companies can enrich unstructured data they possess (accounting for about 90% of the data companies hold). Enriching product master data and product data were given as examples, and by adding information such as brand names and feature tags, generative AI can produce more accurate outputs.

It also showed its effects through actual examples, such as generating product information for e-commerce sites and a new clustering method using generative AI based on ID-POS data.

*Product data refers to product data that is not limited to mere transaction and sales management, but to product data valuable to customers. It includes data that was traditionally overlooked, such as categories, product feature tags, search tags, grouping, and review information. When using Lazuli PDP, you can use external product data collected by Lazuli.

The Importance of a Knowledge Database in Customer Support as Advocated by Givery

Next, Mr. Moriya of Givery took the stage and explained the importance of knowledge databases and how to enrich them in the use of generative AI for customer support. Traditional customer support knowledge (data such as manuals and the experience accumulated in people's heads) had issues such as being dispersed and not updated in a timely manner, so it was not being properly managed or utilized, but the emergence of generative AI showed the possibility of solving these issues.

Mr. Moriya said that by incorporating the concept of KCS (Knowledge Center Service), customer support can be greatly streamlined. Specifically, by continuously updating knowledge during customer support operations and enabling generative AI to automatically generate answers or draft reply messages while always referring to the latest knowledge, the quality of outputs improves and work becomes more efficient.

As a concrete example, we introduced a case in which the knowledge management for customer support at a consumer electronics retailer's physical stores and e-commerce site was integrated, making it possible to provide automatic responses based on response manuals for returns, cancellations, and more.

The Future Brought by Database Enrichment

In the discussion by the two speakers, they talked about the importance of structuring and enriching data scattered across internal departments and websites in order to use generative AI. This enables generative AI to produce outputs without hallucinations and further promotes operational efficiency at companies.

Through this webinar, Lazuli and Givery promised to continue supporting companies that are promoting the use of generative AI by proposing methods for database enrichment.

Company Overview

＜About Lazuli Inc.＞

Lazuli is a startup founded in July 2020. Leveraging advanced AI technology, as a cloud service, it is developing and providing the SaaS "Lazuli PDP" that enables everyone involved in business, such as DX, EC, and marketing professionals, to use product information from around the world. It is working to solve issues with product data that is not smoothly synchronized between companies or even within a company as products are manufactured, distributed, and sold. By integrating and organizing siloed data in manufacturing and retail, it supports DX promotion. It aims to realize a world where people engaged in business can easily access product information.
Company URL：https://corporate.lazuli.ninja/

＜About Givery Co., Ltd.＞

Business content:
・HR Tech business (Track)
・Marketing DX business (DECA)
・Operations DX business (MANA｜Corporate GAI)

Company URL：https://givery.co.jp/

Event introduction service URL：https://deca.marketing/service/cloud/ai-cs/
(Customer service platform 'DECA AI Customer Support' powered by generative AI)