When Faster Delivery No Longer Guarantees Growth: Brian Lin on Restructuring Project Work and Governance at Asia Vista Technology

Asia Vista Technology Co., Ltd. was established in 2002 and provides customised software development, business-process automation and system-integration services. Its public case page lists projects involving banking, insurance, transport and tourism-related systems.

Brian Lin serves as CEO and CISO. According to the company's written confirmation, he holds final responsibility for corporate strategy, project delivery and customer relationships in his CEO capacity. As CISO, he is responsible for approving information-security policy, allocating resources and conducting management review. The AI-assisted development workflow and organisational changes covered by this record fall within those responsibilities.

Asia Vista's system-integration work has long depended on project staffing, domain knowledge and delivery experience. As generative AI and AI Agents entered software-development workflows, the management issue was not limited to whether engineers should use new tools. It also concerned how a delivery model built around labour and time should absorb changes in productivity.

Lin's assessment was that faster development would not necessarily translate into business growth if pricing and organisational structures remained unchanged. This is an attributed executive judgement, not a universal conclusion about the system-integration industry.

Three existing conditions constrained the decision. Financial-client data could not be placed into public models without appropriate controls. Highly customised and legacy-core systems could not be presumed suitable for one standard AI workflow. Organisational change also had to preserve delivery stability for existing customers.

The company therefore did not place comprehensive AI adoption above delivery discipline, system stability or information security.

According to the written response, the judgement developed in three stages. Internal learning about large language models began in late 2022. In 2024, the company tested open-source speech and model tools and used local deployment for selected applications. By mid-2025, it was breaking work involving functional specifications, technical documentation, data models, code generation, testing and operating documentation into repeatable stages.

The decision basis included internal tool tests, the allocation of work across project stages, workforce structure and prior delivery experience in financial and government systems. Most of this evidence consists of internal records and attributed written statements.

On 1 June 2025, Asia Vista introduced a new organisational arrangement. Four project teams were reorganised into three delivery teams, while research-and-development participation was expanded. The CEO directly led a cross-functional AI transformation group of approximately three to five RD and BA personnel. The group was not established as a separate department, and its members continued to work on customer projects.

The decision centred on structuring existing development knowledge into an internal method with identifiable execution and review stages, rather than purchasing a single tool.

Committed resources included direct CEO participation, three to five RD and BA personnel, three delivery teams, AI tools, local computing environments and information-security management work.

The company-provided ASVT Project Skill Package Operating Manual v1.1, dated 4 July 2026, describes nine stages and 29 Skills. The stages cover project initiation and requirement collection, system analysis, system design, programming, integration testing, user acceptance, document delivery, production deployment and maintenance.

Lin's combined executive and information-security responsibilities placed workflow adoption, data-use boundaries and delivery accountability within the same governance context.

According to an ISO/IEC 27001 review result notification dated 4 September 2026, Asia Vista underwent certification audits on 20 August and 3 to 4 September against NF EN ISO/IEC 27001:2023 and ISO/IEC 27001:2022. The notification records that no non-conformity was found during the on-site audit and that the decision-making team granted certification.

The stated certification scope covers information-security maintenance and management for organisation-wide system integration, technology research and development, business marketing, administration, office work areas, the computer room and other ICT infrastructure. This record uses the result only within that stated scope. It does not extend the certification decision to this interview, the performance of the AI-assisted development workflow or other enterprise claims. SNN.TW retains the document as controlled evidence and does not make it available for public download.

The nine-stage workflow structures work previously distributed across individuals and projects into identifiable inputs, outputs and review points. It covers functional specifications, demonstration sites, technical specifications, data models, API specifications, code generation, test lists, document synchronisation, deployment knowledge and analysis of existing programs.

The company states that all three delivery teams have adopted relevant parts of the workflow, although adoption depth varies. The process is therefore in organisational use, but it should not be described as uniformly implemented across every project and stage.

The company reports no consistent reduction in total project duration. It has instead observed a shift in work allocation: more time is spent on system analysis and design, while portions of programming and integration-testing work have decreased. Staffing emphasis has also moved from additional programming support towards more system-analysis work in requirement interviews and information organisation.

The company states that headcount grew from approximately 35 in 2016 to a peak of about 55 and later returned to around 45. It attributes the reduction to resignations, retirement and other natural turnover, and states that it did not implement layoffs, an early-retirement programme or a mass-layoff filing during the stated period. A supplementary internal list covered 41 personnel, with a stated average tenure of 9.37 years; 17 entries showed ten or more years of service. Because the list does not cover the same number as the stated current headcount, this record does not calculate a company-wide percentage of senior personnel.

The available material does not provide a consistent basis for a productivity multiple, efficiency rate or on-time-delivery rate. No quantitative outcome conclusion is therefore presented.

The nine-stage workflow is not yet fully applicable to highly customised projects or legacy-core systems, and the company retains traditional development methods for those situations. Adoption depth also varies among the three delivery teams.

These limitations show that AI has not removed responsibility for requirement definition, domain judgement, information security or quality review. Documentation quality, technology age, data sensitivity and the customer's operating environment continue to determine whether the workflow is suitable.

As of the cutoff date, the evidence supports the existence of an organisational and workflow change. It does not establish that the business-model transformation is complete or that a specific company-wide productivity gain has been achieved.

"If the efficiency dividend created by AI is not converted into a new business model, it becomes pressure to cut jobs. We chose to convert it into a new model."

"Most people ask whether AI will replace engineers. We ask how long a time-and-materials company can survive."

"The most expensive investment in AI transformation is not the tool. It is the owner's time."

"In the AI era, the scarcest people are not those who know how to code, but those who know what should be built."

"The greatest transformation risk is not that the new work fails. It is that the existing work is damaged."

The above quote comes from the interviewee's written response, and the quote and objective facts have been confirmed.

This is a non-paid EXEC demonstration interview initiated through an SNN.TW business invitation. An invitation, document submission or written response does not guarantee publication. The interviewed company may confirm names, titles, dates, figures, occurred events, direct quotations and confidentiality boundaries. SNN.TW retains control over topic selection, title, source use, limitation disclosure, editorial judgement and publication.

Matters that had not occurred by the data cutoff date are neither included nor disclosed in the public record. This record is not legal evidence, an audit, assurance, certification, rating or compliance opinion.