Japan wants to ‘design’ its way out of bureaucracy to reach its AI-first government ambitions

For one of the founding members of Japan’s Digital Agency, Takashi Asanuma, design isn’t just an aesthetic choice but a survival skill for modern governance. 

This was a lesson learnt not through a tech lab, but amidst the chaos of a global crisis.  

Speaking with GovInsider, Asanuma reflects how having witnessed the societal confusion during the early stages of the Covid-19 pandemic made him see design as an urgent, ethical mandate for societies. 

“In a time of genuine crisis, I saw firsthand how outdated public systems and services lacking in user empathy directly and adversely affect the lives and safety of citizens, especially those in the most vulnerable positions,” he explains.  

Notably, the gap between how Japan and its peers handled the pandemic was stark, he observes, highlighting how other nations used design as a strategic lever. 

The UK was prioritising users’ needs over the government’s in designing digital services, while Denmark uses inclusive design to foster public trust and Taiwan embraces an agile, participatory response to co-create solutions with civic tech groups to cope with the pandemic. 

Yet, Japan’s legacy red tape struggled to keep pace.  

Asanuma shares how the crisis marked a career-defining turning point for him to lead a mission to re-engineer the Japanese state. 

He started as the agency’s Chief Design Officer and later became the Vice Minister for Digital Transformation. Currently, he is the Consultant to the Minister for Digital Transformation.  

Since its founding five years ago, the agency has been focusing on the critical building blocks of a digital government. It scaled the national digital ID, migrated to the government cloud, and deployed a secure GPT tool for civil servants. 

As Japan sets its sight this year to become an artificial intelligence (AI)-first government, Asanuma shares how the tech alone isn’t the finishing line, but in the total redesign of legacy systems and processes to make AI actually count for the people. 

What entails an AI-first government  

According to Asanuma, an AI-first government isn’t about adding AI on top of existing, dysfunctional systems and services. 

While AI is good for processing large volumes of data, whether right or wrong data, the quality of outcomes is only best achieved through a human-centered design approach, he says. 

In his private sector days, he learnt that to improve user experience, one needs to “go beyond the surface layer and redesign the operating system (OS) that runs society itself.” 

In the govtech context, this entails moving from sporadic AI pilots to the agency taking on the task of establishing shared AI infrastructure across the government. 

Asanuma elaborates on the Government AI policy, which establishes a secure environment alongside the governance mechanisms to support civil servants in using AI.  

The policy entails setting up common applications, data development, and specialised operational services. 

The agency is translating this policy into practice by starting small with quick wins to showcase impact, and then encourage wider AI implementation among civil servants. 

“Since radical restructuring of operations is difficult to achieve from the outset, we are currently focusing on application testing tailored to administrative work. 

“This includes features such as support for organising public comments and searching parliamentary responses, with the aim of improving the efficiency of government employees,” he explains. 

For Asanuma, tech success depends less on the launch itself, and more of adapting underlying processes and systems to make the implementation stick. 

“Issues in administrative experience often arise not from what appears on the screen, but from the underlying institutional frameworks, operations, and systems that shape it,” he observes from his experience. 

These systemic challenges may range from fragmented data foundations, siloed organisational structures, to outdated legal frameworks. 

Why redesign existing systems  

Tech can scale transformation, but not the quality of outcomes which can only be achieved by adopting a user-centred perspective in its implementation, says Asanuma. 

Tasked to implement “human-friendly digitalisation that ensures no one is left behind,” the agency has embarked on initiatives including reviewing major services, recruiting specialised personnel, developing evaluation processes, and more.  

He illustrates how the agency adopted a design approach to implement Mynaportal, which is the government’s online services portal. 

Rolling this out in phases, Mynaportal first went out with a Beta version, allowing users to choose between using the new interface and the old one. 

The official version was later deployed with continuous improvements made based on user feedback and usage monitoring.  

Now onto the next leap of AI, he suggests starting with the users themselves when considering AI implementation.   

Specifically, it is to take the perspective of the people who receive and experience the policy by “understanding their requests, expectations, difficulties, potential misunderstandings as well as anxieties.” 

Asanuma envisions AI to support the government in becoming “a familiar and accessible presence” for citizens, where they can use services tailored to their needs at ease and anytime. 

Part of the agency’s efforts to build public trust is also through transparency, as it publishes annual reports around the citizens’ usage and satisfaction levels of using digital government services. 

The agency is taking the same iterative approach for AI governance as well.

“It is essential to simultaneously develop advanced risk management mechanisms, such as rules for procurement and utilisation,” he explains, highlighting this as a way for the government to build a track record step by step. 

“The design approach is not merely adding value for AI implementation; it is an essential foundation, process, and culture for scaling AI safely,” he says. 

Enabling the cultural change among bureaucrats 

According to Asanuma, civil servants who possess multiple areas of expertise are key toto drive an AI-driven administration, where “policy, legislation, operations, data and systems, security and experience design operate in an interconnected manner.” 

To cultivate this multidisciplinary skillset within the sector, he shares that the agency has been developing collaboration frameworks between technical (product managers, designers, engineers and architects) and administrative civil servants. 

He adds that process design matters as much as the product design itself. 

“This includes agreeing on policy objectives and value judgments, introducing decision-making processes with multiple layers of review, and establishing clearly documented feedback processes,” he explains. 

How far should the public sector go in its pursuit of experimentation? Asanuma says that the question isn’t whether the government should experiment, but how it can do so without breaking public trust. 

By establishing “safe learning environments” from the beginning, the public sector can transform risk into a controlled lab. This also entails the following elements: 

• Starting small by limiting the scope, duration, and functionality 

• Capping impact by designing safeguards that limit maximum damage  

• Defining the conditions and metrics on when to stop 

• Establishing clear pathways for accountability by communicating objectives, results and subsequent improvements 

He believes that combining the elements sends a clear message to civil servants that experiment doesn’t equate to irresponsibility, but an improvement cycle enabled by transparency and accountability.  

Closing the conversation, Asanuma reflects on the designers from various sectors around the world* who have inspired his work in the public sector. 

“Designing for society requires blending empathy for individual pain with an ethical imperative to leave no one behind. All while maintaining the pursuit of simplicity and transparency,” he stresses. 

Asanuma stresses “good design” in the government is more than making services easy to use, but building a trustworthy and inclusive system that empowers every citizen to participate in solving collective challenges. 

*Asanuma shared that he has been influenced by Dieter Rams’ emphasis on both honesty and beauty in design, Don Norman’s idea of human-centered empathy, and Victor Papanek’s highlighting the social responsibility of design to respond to human needs. 

To read past coverage of Japan's Digital Agency, please click here to go to our Global Digital Government Directory.