Issue
・The criteria for business decisions based on internal documents are scattered across multiple materials and individual experiences, resulting in variability in response quality.
・Less experienced members and transferees lack sufficient decision-making materials, leading to frequent confirmations with veteran employees.
・There are no unified rules for document structure or writing style, causing feedback such as "hard to understand" and "difficult to find information."
The operation depended on individuals for judgment criteria and writing style, and a reproducible workflow had not been established.
Support Details
At Human Science, we provided support for manual standardization, improvement of writing skills, and utilization of generative AI.
1. Manual Standardization Support
・Inventory of existing materials and analysis of inconsistencies and difficulties in understanding
・Creation of model manuals and templates
→ Establishment of a system that ensures the same structure and quality regardless of who writes it
2. Technical Writing Training
・Conducted feedback-based training using practical documents
・Learned methods for organizing information, designing structure, and expressing sentences
3. Lecture on Utilizing Generative AI
・Shared concepts for utilizing generative AI
・Designed a writing workflow for authoring manuals using generative AI
・Created prompt templates for authoring manuals using generative AI
Results
After the training, a common language for technical writing was established within the company, improving document quality and decision accuracy.
・Completed the first edition of the manual with high quality
・Reduced number of inquiries
・Reduced workload for document creation and review
Such effects have been observed.
Through this initiative, manuals have evolved from "something that is created" to "something that is used on-site." Additionally, Company A utilized generative AI not only for the writing tasks but also for researching the knowledge necessary for the target operations. As a result, they established a process that is not reliant on the individual who "knows the operations well writing the manuals," but rather allows "people with limited operational knowledge to participate in the writing process, supplementing knowledge with generative AI to maintain a certain quality."
