Ajira Co., Ltd. 

• Incident Lake

• Professional Services

Azira Inc. is a startup that develops AI products primarily for the security industry.As the company experienced rapid business growth, it faced operational challenges such as the siloing of incident response knowledge across teams and high communication costs in a multilingual environment. To address these challenges, the company implemented SIGQ’s Incident Lake and professional services. This enabled the establishment of a cross-team knowledge-sharing platform and the creation of a smooth incident response system that transcends language barriers.

■Challenges Before Implementation

• Knowledge regarding incident response and system monitoring tended to be siloed across different business and product teams

• The overhead caused by the language barrier between Japan and Vietnam had become a bottleneck in development and incident response.

• We were aiming to strengthen our operational structure and improve operational quality in order to further scale our business

■Benefits of Implementation

• Use Incident Lake as a shared platform across teams to seamlessly share incident-related knowledge

• AI instantly shares analysis results and discussion content in multiple languages, reducing the effort required for translation and minimizing misunderstandings caused by language barriers, thereby facilitating smoother communication.

• By leveraging operational insights from our professional services, we can scale your business further, including overseas operations

Transforming the Security Industry with Behavior Recognition AI

Azila Inc. is an AI startup founded in 2015 by a joint team of members from Japan and Vietnam. Leveraging its proprietary behavior recognition AI powered by deep learning, the company develops both in-house and custom AI-driven products.

One of the company’s core businesses is “AI Security Asilla,” an AI product designed for the security industry. The security industry continues to face a chronic labor shortage. Consequently, the task of monitoring camera footage to detect anomalies was an area that was highly compatible with the company’s behavior recognition AI. AI Security Asilla is highly regarded for its low implementation costs and ease of deployment, as it can be set up simply by connecting an edge AI server to existing security cameras.Additionally, the AI analysis of camera footage is performed entirely on the edge, a key feature that enables advanced behavior recognition while minimizing security concerns. The cloud infrastructure the company built on AWS is responsible for managing these edge AI servers. Specifically, it handles tasks such as health monitoring, software updates, and the collection of operational logs.

However, as the business expanded rapidly, several operational challenges began to surface at the company.

The primary challenge was the siloed nature of incident response across teams. While both the product development team and the contract development team had accumulated their own operational know-how, this information was not shared smoothly between teams, leading to challenges in reusing knowledge and standardizing quality. For example, while the contract development team excelled at AI operations and the product team excelled at uptime monitoring, information sharing between the two was reportedly insufficient.

The second challenge is the increased communication costs resulting from language barriers. The company has adopted a division of labor in which its Vietnamese subsidiary handles the majority of development, while the Japanese subsidiary is responsible for planning, design, and some operational tasks. As the product and organization grew, the overhead associated with communication across multiple languages became impossible to ignore and was beginning to act as a bottleneck to scaling operations.

The third challenge was strengthening our operational framework. Given the nature of our product—which is designed for the security industry—it requires a high level of reliability, yet the operation of edge AI servers presents its own unique complexities. We had to address hardware-related issues, such as camera streaming problems, as well as AI-specific issues, such as AI malfunctions and missed anomaly detections. Consequently, even though the edge servers were centrally managed on the cloud side, many tasks still relied heavily on manual intervention by engineers.To further expand the business in the future, it was necessary to centralize the management of these incidents while aiming to transform the operational framework through more advanced automation and improvements in the quality of deployments and releases.

Looking back on that time, the company’s CTO, Masahiro Wakasa, summarized the challenges the company faced, stating, “We must advance automation as much as possible and evolve into an operational organization where humans can make decisions quickly.”

Breaking Down Silos at Incident Lake: The Power of Knowledge Sharing and Multilingual Support

To address these challenges, Ajila implemented SIGQ’s Incident Lake and Professional Services.

Incident Lake is an AI agent that consolidates incident-related operational data scattered across the organization to support incident response. The company uses Incident Lake as a common platform to facilitate the sharing of knowledge across multiple business units and product lines.

Specifically, the company consolidated operational knowledge regarding edge AI servers and incident information from the application layer into Incident Lake, enabling centralized storage and access across departments. Behavior recognition AI and edge operations are core areas of the company’s technology, and there were many aspects that could be standardized across multiple teams. Mr. Wakasa notes that the ability to share this knowledge across departments has greatly contributed to breaking down silos.

Furthermore, Incident Lake’s multilingual capabilities play a major role in reducing communication costs. By integrating AI into the incident management tool, it prevents the loss of subtle nuances that often occurs in multilingual communication and ensures the accuracy of information. As a result, Mr. Wakasa notes that team members can now focus more on discussions and decision-making without having to spend time on translation or interpretation.

In addition, Mr. Wakasa finds the automation of knowledge management provided by Incident Lake particularly beneficial.Manually creating tickets and entering knowledge across various tools had become a significant burden. By leveraging Incident Lake, the team can now automatically accumulate flow information from sources like Slack as stock information. Mr. Wakasa highly values the fact that this automation of knowledge accumulation allows for implementation without the need to spend time mastering new tools or altering existing operational workflows.

“Incident Lake is designed with a user experience that minimizes the need for manual data entry. As a result, we were able to integrate it seamlessly into our operational workflow while minimizing resistance among team members to the new tool,” said Mr. Wakasa.

"Will They Really Go This Far?" The Path to Operational Transformation Support That Works Side-by-Side with the Front Lines

Along with Incident Lake, Azila also implemented SIGQ’s professional services. For the company, which was looking to strengthen its operational capabilities, the hands-on support provided by SIGQ’s experts came at just the right time.

SIGQ’s support begins with a thorough assessment of the current situation. In this company’s case, we first conducted an inventory of the location and usage status of data scattered across various internal tools, such as Jira and Slack, to lay the groundwork for centralized data management.Furthermore, the team visited the development office in Hanoi, Vietnam, and through discussions with local members, identified challenges and communication issues that were not apparent from documentation alone. By organizing the challenges within the current operational framework in this way, the team is working to transform the operational organization in collaboration with on-site members.

We have also begun developing solutions to support organizational transformation, including an API that automatically creates tickets in Incident Lake from our in-house operations tools, as well as designing data integration between Incident Lake and various operations and communication tools. Furthermore, we envision a future where we combine the operational logs accumulated in Incident Lake with AI-driven development to achieve end-to-end automation—from incident detection and resolution to build and release.

Mr. Wakasa says he has great confidence in SIGQ’s ability to support operational transformation.

“I was honestly surprised that they got involved on the ground even before the official support began and offered various suggestions for improvement—I thought, ‘Are they really going to go this far right from the start?’ Thanks to their thorough support, I became convinced that this service is essential for strengthening our operational structure,” said Mr. Wakasa.

Leveraging our operational expertise to bring AI-powered security to the world

Regarding the company’s future outlook, Mr. Wakasa stated that behavior recognition is a highly versatile technology that can be applied regardless of country or region, and that the company is moving forward with preparations with an eye toward expanding its business overseas. At the same time, Mr. Wakasa noted that strengthening operational capabilities is essential for the smooth expansion of overseas operations, and he announced that the company will continue its efforts to transform operations using Incident Lake as a starting point.

"As our customer base grows, we need to raise the bar even higher for operational quality. We aim to build on the insights gained from our professional services and Incident Lake to deliver even more reliable services," said Mr. Wakasa.

Finally, Mr. Wakasa offers the following message to companies facing similar challenges.

“Every company faces operational challenges. Furthermore, there is a limited pool of talent with extensive operational expertise, and the reality is that recruiting such individuals is extremely difficult. Given this situation, I find it very appealing that leveraging professional support can shorten the time required to resolve issues and quickly improve operational quality,” said Mr. Wakasa.

Azila is pioneering a new market in AI-powered security. SIGQ will continue to support the company in strengthening its operational infrastructure and expanding its business globally through incident management.