Gaming / Game development
COLOPL: Company-wide generative AI use and work automation tools built by employees
- Company
- COLOPL
- Country
- Japan
- Adoption stage
- In operation
- Source published
- Date basis
- The date the source was published. It can differ from the date adoption started.
- How the source was checked
- Read the full source text
The work problem
According to an official COLOPL announcement, the company uses AI to give users better entertainment on the premise that people are always the main creators. Employee comments in the same announcement show that a lot of time went into creating placement data for game stages, implementing new features including QA, and finding the right files among large numbers of CSV and PDF files. According to the announcement, small automation tools for work efficiency also used to require help from engineers.
Technology and data
According to the same announcement, COLOPL set up the AI Enablement Group in April 2025 as a specialist unit under the CIO office, and the group leads the adoption of the latest AI tools, guidance on how to use them, and systems for sharing knowledge inside the company. The HR strategy department holds a weekly AI knowledge sharing session for back office staff, where employees present cases in which they automated their own work and the trial and error involved, and the technical PR team regularly runs AI study sessions for engineers on specialist topics such as coding and agent development frameworks. Employees build their own work automation mechanisms such as Gems, skills and bots, and the announcement gave examples of designers now building small automation tools themselves and an employee building a web tool that brings together separately managed spreadsheets for projects, sales and business partners. The announcement stated that people decide to the end what feels fun and what experience to deliver to customers. Employee comments say that people adjust the fine details of stage layouts made with AI by hand, and that new features are built with vibe coding and played to confirm they are fun before engineers are asked to implement them.
Results
According to the same announcement, 81.1% of respondents in an internal survey of all employees said they felt AI use had reduced their working hours. Based on weekly time savings self-reported by the 343 employees who said their hours fell, excluding one outlier and counting employees who felt no reduction as zero hours, the company estimated that the average across all 423 valid responses converts to about 25.8 hours per person per month. 78.5% said the quality of finding work problems, planning and generating ideas had improved, and 91.6% of them said this had led to more accurate execution and production. 52.7% said they had built work automation mechanisms themselves, and the counts self-reported by the 223 employees who said they had built one add up to 1,091. Employee comments include examples of cutting the time to create stages to about one fiftieth, shortening a new feature implementation estimated at 15 business days including QA to about 5 business days, and cutting a file search that took about 2 hours to 5 minutes.
Limits and open questions
All figures come from an internal survey answered by employees themselves, and about 25.8 hours a month is an estimate converted to a monthly figure from self-reported values. The announcement gave the survey population as all 566 employees as of September 2025, and did not disclose when the survey was conducted, the names of the AI models and tools used, how much the 1,091 automation mechanisms are actually used, or how the saved time affected development periods or sales.
Sources
- 8割が業務時間短縮を実感、現場発の自動化ツールは1,000個以上 ～AIで生まれた“余白”をより良いエンタメづくりへ～colopl.co.jp, Accessed
Compiled from public sources. These are not results from ATF Works customers.