Gen-Z Team Creates AI Platform ‘Bureaucrazy’ to Make Government Work Easier
मंसिर १, २०८२ १२:५३
Kathmandu: Aakriti Ghimire, who works closely in government policy-making, was always troubled by one question, why is Nepal’s bureaucracy not simple and people-oriented? Why is it constantly criticized? Why do people feel uncomfortable when they visit government offices?
She also worked in the secretariat of then Education Minister Sumana Shrestha and searched for a solution. During that time, she realized that the main weaknesses in the bureaucracy were complex legal provisions, procedural hassles and a lack of policy information. She focused on how to solve them.
“The young generation today is lagging behind in politics, policy-making and bureaucracy due to the lack of government information in a language we understand. The lack of simple information has become the main reason for the bureaucracy’s disrepute,” Aakriti says. “I thought that if information could be provided in a simple way, this problem would be solved to some extent.”
Aakriti reached Incubate Nepal four months ago with the same problem and the solution, Bureaucrazy.co. Incubate Nepal is a virtual eight-week program. It teaches students skills and builds an environment for collaboration between students and people working on different projects.
During the program, Aakriti mentored Manish Singh Mahato, Drishya Puri, Prashim Timsina, Sarvashree Subedi and Aakrishta Dahal, who were learning programming after completing class 12. Prisha Bajracharya, Preeti Pant, Niharika Chapagain, Kriti Thakur and Rishavi Gautam joined as fellows. Aakriti had to explain her concept to 10 students, and the students had to develop it into a product.
Data scientist and lecturer Lobi Karmacharya and e-commerce professional Ayusha Shrestha also joined the project to mentor students.
How Was Bureaucrazy Created In Eight Weeks?
Aakriti explained the main concept to the team. Aayush and Lobi helped turn it into reality. “Legal data and procedures are in the respective offices, but they are not organized. They are not easily accessible. They are not integrated. Even if they are, not everyone can find them. Our goal was to show that all these can be integrated and made available on a single platform,” Aakriti says.
In the initial phase, students and mentors collected necessary documents and interviewed representatives of government offices. They planned to include all the verified data in the platform.

In the first week, all participants learned about the project’s objective. Some shared ideas, some worked on the process, and others pushed the platform forward. In the second week, they held regular discussions.
They studied the law and talked about how the law shapes the NOC process. They learned how to simplify the complex language of the law. In the third week, the development of the Bureaucrazy platform moved ahead.
After building the platform in the fourth week, they prepared to make it public. Bureaucrazy’s social media accounts also became active. Work to make the chatbot effective continued.

In the fifth week, they worked to make the chatbot more standard. In the sixth week, they focused on strategy. They worked on making the chatbot and website accessible. They interviewed people who had taken or given NOC and gathered experiences. They also met government employees to discuss simplifying the process. In the seventh week, they completed the FAQ. They refined the landing page. In the eighth week, they gave the final presentation.
“The platform was ready in just eight weeks,” Aakriti recalls. According to student Manish Mahato, technically, only the domain has cost money so far. “We paid for the domain,” Manish says. “We have used open source platforms for everything else.”
He says the current chatbot is based on the Retrieval Augmented Generation model. “We have not trained an LLM,” Mahato says. “We extract data ourselves by verifying it with government guidelines, laws and bureaucrats and convert it into vector embeddings.”
Explaining simply, Lobi says, “Usually, websites have FAQs. That is the initial concept of this platform. We have limited verified data and expert conversations. After this, the chatbot tries to understand and answer whatever the user asks.”
According to Lobi, when a user asks a question on the platform, the AI searches the database and answers in simple language. In the initial phase, it provides information on NOC (No Objection Certificate), Letter of Evaluation (LOE) and election-related procedures.
The Path After Becoming The Runner-up
No one had planned how to take Bureaucrazy forward. Aakriti, Ayusha and Lobi had taken time from their main jobs to mentor the project. So they could not devote full time.

Meanwhile, they learned about the Kevin Zhu Innovation Challenge organized by the Rhodes Forum on Technology and Society under Oxford and applied. During the month-long program, mentors provided continuous guidance. They presented a project and model on how AI systems can support bureaucratic work. Their project was recognized as excellent and on 30 September 2025, they became the runner-up. They also received a cash prize. (The team did not want to disclose the amount.)

After winning, they received financial support to advance the project. And Bureaucrazy, which was limited to its eight-week beginning, continued.
“Even though the product was ready, it was just a vision,” Aakriti says. “We used mostly free open source software. We were unsure how to take it forward. After winning the award, we had to make good use of it. Now we are making it more standard by using premium software.”
The platform is still in its initial phase. “There are no experts involved yet. Technically, all roles are played by students who completed class 12. Now we are moving to make it standard,” she says.
Purpose Changed After The Gen-Z uprising
Initially, the platform aimed to inform students about NOC and LOE. But the Gen-Z uprising on September 8 and 9 made them realize they should also inform the public about other aspects of bureaucracy. They felt they should make the public aware of the upcoming March 5 elections and the process required to get a voter ID card.
“Initially, we thought of a few limited procedural things,” Aakriti says. “Now we realized we should provide information about the bureaucracy and the form of governance. For example, election-related knowledge such as the structure of government, the form of governance, and what proportional and direct systems mean.”
Plans And Challenges Ahead
The AI chatbot on Bureaucrazy.com can now understand NOC, NOE and voter ID card queries. According to mentor Lobi, the AI chatbot can be asked questions in Nepali, English or Roman Nepali. It provides information based on the resources available.
“Now, no matter how the question is asked, it analyzes it and answers in English,” Lobi says. “We will soon answer in Roman or Nepali Devanagari as well.” It also has a digital ‘Law Library’. Although the resources are limited now, they plan to expand it.
Aakriti says they have also begun providing information about how students can study law, what free resources exist and other materials.
The website also has sample letters used in government offices. These samples can be downloaded and used.

The primary goal now is to make Bureaucrazy technically strong. The team aims to fully train an LLM by enhancing the chatbot’s capabilities. They are also working on a voice recognition feature so smartphone users can ask questions by speaking.
To expand the platform, they will hire legal interns and add information on public bureaucratic processes such as land revenue, taxes, licenses and citizenship.
Another major goal is transparency. Each answer given by the chatbot cites its source. They say it will be based on certified documents published by government bodies or expert opinions.
They believe this will maintain the credibility of the information and control AI hallucinations. According to Lobi, the main challenge now is building the capacity to serve millions of users at once and shifting from free software to a reliable paid system.
The long-term vision of Bureaucrazy is to improve the relationship between citizens and the bureaucracy. Mentor Aakriti Ghimire says this project is a social initiative, not a business model.
Believing that “information is power,” the team says this platform can help control corruption by ensuring easy access to information. Often, due to a lack of information, people pay informally to expedite services or understand procedures.
They believe the platform will help stop this trend. It will also build confidence among citizens that they can solve their own problems. After the project’s success, they aim to integrate it as a plug-in chatbot on federal or municipal websites in the future. However, due to uncertainty about a systematic and sustainable business model, questions remain about its long-term presence.
पछिल्लो अध्यावधिक: मंसिर १, २०८२ १२:५३
