Generative AI: What Business Leaders (Even Those from Liberal Arts Backgrounds) Should Know

I'm Kunihiko Monbun, CEO of GNUS Inc.

This series, titled "What You Need to Know About Digital Products Today," focuses on DX and digital products to help Japanese companies advance their DX efforts.
This fourth installment discusses how to incorporate generative AI.

GNUS Inc.: Established within the Dentsu Group in 2019 as a partner for business growth and transformation through digital products. We support clients from planning and PoC through development, operation, and growth of digital products—key to DX for new and existing businesses—by assembling optimal teams from our network of over 600 members globally and applying agile project management.

Four Steps to Utilizing Generative AI

Generative AI is rapidly becoming an indispensable tool in the business world. Its applications span from simple chatbots to applications that dramatically improve operational efficiency.

Here, I'll share the four steps I'm personally using to implement generative AI incrementally. Following these steps should help you unlock its true value and focus on tasks requiring greater creativity than the AI itself. For context, I have no programming background—I'm a humanities graduate. Yet, achieving this level requires almost no coding.

The first step in utilizing generative AI is implementing it as a chatbot. It can quickly verify information you need to look up during daily tasks, draft documents, review written text, or summarize content.
For example, when I needed to create a simple SQL query (a programming language for storing and processing information in relational databases) in the past, I used a generative AI-powered chatbot to quickly build it. I also use it to check for inconsistencies in wording and expressions in documents I've written.

Key points:

• Streamlining simple tasks... For example, tasks like refining emails or articles, translation, and checking for inconsistent expressions are relatively easy to automate, so starting here is a good idea.
• Generation beyond text... It's not limited to text; generating images and videos is also straightforward. Image generation, in particular, can easily be utilized for slides or posted on websites.

 
Once you've experienced the potential of generative AI through chatbot implementation, the next step is prompt tuning. By refining prompts, generative AI evolves from a simple chatbot into an agent that handles routine tasks. For example, it can automate creating sales reports, summarizing meeting minutes, or replying to emails.
Prompt tuning is especially useful for creating articles like columns. Rather than asking it to write everything from scratch, simply adding the overall structure and headings to the prompt, along with instructions on length and formatting, can yield satisfying results.

Key Point:

• Prompt Engineering... Prompt engineering is the technique of providing appropriate instructions to generative AI. Learning prompt engineering allows you to maximize the capabilities of generative AI.
• Reviewing Business Processes... Before introducing generative AI, it's crucial to review existing business processes. Designing processes suited for generative AI enables more efficient workflows.

 
As generative AI adoption advances further, you can develop applications that incorporate external information and internal proprietary data to streamline more sophisticated tasks. For example, tools can be developed to collect and analyze competitor information, predict market trends, or analyze customer feedback.

This step actually involves various use cases. In my application, I feed it materials like corporate earnings conference presentations to generate summaries or compile points of interest. It also easily checks for differences and risks between our company's contract templates and those proposed by clients.

Key Point:

• Leveraging RAG*: By incorporating proprietary company information (such as internal policies and operational data) not used in training generative AI models, you can create chatbots optimized for your specific operations or utilize them as business applications.
• Leveraging Fine-Tuning... Combining RAG with fine-tuning simplifies prompts, improving cost efficiency and response speed.

※Retrieval-Augmented Generation. A technology that improves response accuracy by combining text generation using LLM (language models built with massive text data and deep learning) with external information retrieval.

 
The final stage of AI utilization is integration with internal systems. By connecting generative AI with core systems, CRM tools, SFA tools, etc., applications can be developed to streamline workflows for multiple team members. Examples include apps that automatically visualize sales activity progress, optimize customer outreach, or enhance internal operations.

At GNUS Inc., we deploy several internal applications for staffing operations. We've created an app that converts project staffing requirements from free-form text into a standardized format
, which has helped improve both the speed and quality of staffing.

Key Point:

• Operational Framework... When introducing applications utilizing generative AI, establishing an operational framework is also crucial. Designating personnel responsible for application maintenance and operation, and performing regular maintenance and updates, helps maintain stable operation.
• Security……When integrating with internal systems, robust security measures are vital. Ensure the confidentiality of data handled by generative AI and implement safeguards against unauthorized access and information leaks.

Summary

Generative AI is a powerful tool with the potential to revolutionize business. Even business leaders with a humanities background can achieve operational efficiency, productivity gains, and the creation of new business opportunities by gradually leveraging generative AI, using the four steps introduced in this article as a reference.

We would greatly appreciate hearing how you are utilizing generative AI and where you encounter challenges.

Kunihiko Bunbun / CEO, GNUS Inc.
After joining Dentsu Inc., he was assigned to the Sales Division, where he worked on marketing strategies for foreign consumer goods manufacturers and IT companies. In 2009, as part of Dentsu Inc.'s new business division, he launched the digital magazine sales app Magastore and served as its Product Manager. From 2011 onward, within Dentsu Inc.'s newly established New Business Development & Consulting Division, he promoted digital transformation and new business consulting for the media, finance, automotive, and sports business industries. From 2017, seconded to Dentsu Holdings USA in New York, primarily responsible for digital marketing consulting and new business planning for major Japanese manufacturers, driving the development of software services utilizing AI. Returned to Japan in 2019, founded GNUS Inc., and assumed the position of CEO.