Development speed becomes an issue when developing products for our own brand

As a "technology trading company," Macnica has added technological value to products such as the world's latest semiconductors, electronic devices, network-related equipment, and software and provided them to customers. Currently, we are actively working on developing our own brand products with the aim of transforming into a service solution company. Among them, the Innovation Strategy Business Headquarters, which was established by commercializing projects such as IoT, AI, DX, etc., has already released many AI-related services, such as releasing the AI-related service "Re: Alize. macnica. ai" in 2020. It has a proven track record and knowledge, and has achieved the number one market share in multiple sectors such as manufacturing and construction in the image recognition market. Osamu Mori, Product Planning and Development Section Manager, Digital Business Development Department, Innovation Strategy Business Headquarters, said, ``This year in particular, we are focusing on ``creating first'' in DX product development.Recently, we announced a product called ``Omatome Ninja.'' We are actively developing service solutions such as this."

Until now, the company had often done offshore development by outsourcing it to partners in India, so there were not many opportunities to manage the source code. However, as in-house development increased, it became necessary to properly manage the source code by ourselves. Therefore, they used GitHub Enterprise Server, which was used by another development team, but problems arose in the speed of development.

“The products we develop incorporate many AI technologies, but the problem is that it is difficult to guarantee their accuracy.In general development, it is possible to predict results based on specific logic. However, when AI is used, it tends to be difficult to predict.Furthermore, because users have little experience using AI in practice, they are unable to judge how their work will change when a new product is introduced. (Mori)

Therefore, we proceeded with product development by having users actually touch the product and making improvements while receiving feedback and requests. In this type of development method, how quickly improvements can be reflected and commercialized into products, in other words, release speed and frequency are important. Until then, the company had been using GitHub Enterprise Server as its version control system, but there were challenges in speedily improving and releasing products. Therefore, the company decided to consider migrating to other products, including GitHub EnterpriseCloud.

GitHub Enterprise Cloud was selected based on the ease of collaboration between development members and the AI coding support function.

In the existing development environment, GitHub Enterprise Server was set up on the internal network, and a pipeline was created using GitHub Actions' Self-hosted Runner to deploy to the production environment server on AWS. However, it was necessary to check that the configured pipeline was running correctly and to maintain the self-hosted runner, which was a burden on the maintenance work and the time it took.

``When introducing a new development environment, we considered two products in addition to the GitHub Enterprise Cloud that we adopted this time.GitHub Enterprise Cloud is a product that the Company handles, but we are only evaluating its advantages and disadvantages from a pure user perspective. We compared them.” (Mori)

Of the two products that were compared, we felt that the product from a certain major vendor lacked collaboration functions between development members, including code review, and that at the time of the study in 2022, coding support using generative AI was lacking. It was dropped because the implementation of the function was undecided. The other product, which is also used by offshore partners, is open source and costs less, but it lacks technical support and maintenance must be carried out by the company itself. Also, this product similarly did not have an AI coding function. In the end, the company decided to use GitHub Enterprise Cloud because it had all the necessary features and was easy to use.

Migration to GitHub Enterprise Cloud significantly improved development speed and operational load, and facilitated collaboration with external development partners.

The first effect of implementing GitHub Enterprise Cloud is that development speed has improved significantly. In the first place, when we switched from offshore development for an Indian company to in-house development using GitHub Enterprise Server, the time spent communicating with the local people was eliminated, and the time it took to develop something that could be shown to the customer as a result was 3 months or more. It was shortened from half a year to about one month. Furthermore, by migrating to GitHub Enterprise Cloud, the time to release has been further shortened. It has improved to the point where it can provide weekly updates.

Collaboration with external development partners has also become smoother.

"When we were using GitHub Enterprise Server, we had to arrange for dedicated PCs that could be connected to our internal network and provide them to partner companies. It takes a considerable amount of time from arranging them to distributing them, so we are now in a situation where we are in a situation where we need more manpower. However, by moving to GitHub Enterprise Cloud, we were able to grant access rights to the PCs of individual development partners outside the company, so it was difficult to respond in a timely manner even when the company needed a PC. This eliminates the need to arrange for new features.As a result, by granting access to the target repository on GitHub to the target user, they can immediately engage in development.Also, since it is a cloud service, new features can be added to the target user immediately. Another advantage is that it can be used as soon as it is released.It's nice to be able to try out new features of code generation AI right away.'' (Mori)

In fact, by using the coding support function (GitHub Copilot) using generative AI, it became possible to quickly create and review code, and the speed of development was significantly improved. Currently, each developer is searching for the optimal method through trial and error, how to utilize GitHub Copilot to generate high-quality code, and how to use it to lead to efficient development. That's what it means.

There are also significant operational benefits. Migrating from GitHub Enterprise Server to GitHub Enterprise Cloud has made it easier to deploy to production environments. Furthermore, there are a wide range of collaboration tools between GitHub and AWS, and by using these tools, it is possible to deploy without the CI/CD server that has been used up until now, and monitoring and maintenance of Git servers and CI/CD servers is no longer necessary. .

"When we were using GitHub Enterprise Server, the CI/CD server would go down once every few months, and it could take an entire day to recover. As a result, the schedule, release speed, and productivity of the development team members were affected. There were some negative effects.Furthermore, GitHub Enterprise Server requires periodic version updates and patching, and each time you have to worry about whether it will start up without any problems.After migrating to GitHub Enterprise Cloud As a result, that risk has completely disappeared.'' (Mori)

GitHub is recommended for users who want to develop efficiently while properly managing source code.

Regarding the future, the Innovation Strategy Business Headquarters is considering further utilization of coding support functions and application security functions.

“We recently released GitHub Copilot Chat, a feature of GitHub Copilot Business that allows you to interact with GitHub Copilot in the editor and receive real-time guidance on your source code. I would also like to incorporate GitHub Advanced Security, which detects security risks hidden in code early, so that I can improve release speed while maintaining high quality.Recently, we are producing software for our own products in-house. , an increasing number of companies are moving toward in-house development, and GitHub is a tool recommended for users who want to manage their source code safely and appropriately within their own company.In addition, for development teams of new businesses, I think there are many cases where you are asked to jump into a field in which you have little experience and develop.Even in such cases, receiving support from GitHub Copilot lowers the hurdles for developing something you are not familiar with, making it easier for you to get involved in a new business. This is a tool that is recommended for those who are.” (Mori)