Digital memorial, AI voice bot to practise mother tongue among new tech trialled in Singapore

SINGAPORE – Growing up, Mr Liau Wen Rui learnt about the life of his fisherman grandfather, who died before he was born, through stories from his relatives.

“When I look at his niche during visits (to the columbarium), all I see is his name, photo, date of birth and death,” said the 29-year-old, who is a senior product manager at the Government Technology Agency of Singapore (GovTech).

“I always thought, there has to be a better way to capture memories of him for myself, and for future generations of the family.”

So when the inaugural {build} hackathon organised by GovTech Singapore rolled around in February, he partnered four colleagues and the National Environment Agency to build AfterLifeSG, a digital platform that allows relatives and friends to contribute memories of a loved one.

The platform allows users to create memorials of a dead family member, share a personal description of them and provide the location of their niche or grave plot.

Relatives and friends can then access the website to leave virtual flowers, and share memories and photos of the person. Users need to log in via Singpass – an intentional step to create a safe sharing environment and prevent anonymity, said Mr Liau.

“I often wonder how remembrance in the future will look like for people around my age,” said Mr Liau, who added that in land-constrained Singapore, remembrance facilities such as Bukit Brown Cemetery have been repurposed.

The platform is currently in a pilot phase for employees of all government agencies, to allow the team to gather feedback and iron out back-end processes such as the verification of death certificates.

The eventual aim is to roll out the platform for public use in future, Mr Liau said.

The team is also considering ways to make the process of paying respects digitally more specific to local cultures.

AfterLifeSG is one of more than 20 tech prototypes on display at the 2025 GovTech Innovation Day, held on Nov 13 at the Sands Expo and Convention Centre.

In her opening address, Minister of State for Digital Development and Information Jasmin Lau urged employees of GovTech and other government agencies in the audience to help modernise Singapore’s legacy systems.

Around one-third of government systems will be more than 10 years old by 2027, said Ms Lau, adding that some have become a bottleneck in how services are delivered to citizens.

It is not just replacing old with new – but creating systems that are flexible, scalable and built to evolve, while also embedding robust digital security measures to ensure resilience, said Ms Lau.

“We want to make sure that every public servant – from front-line workers to senior leaders – has the skills and mindset to use digital technologies to serve Singaporeans.”

She added that permanent secretaries have undergone training sessions to understand the principles of product development, system modernisation, and the potential and risks of artificial intelligence (AI).

“Our ministers will (also) go through training sessions designed specifically for them... I invite all of you to join us on our learning journey.”

Another project on display was LangBuddy, a web-based AI voice-enabled chatbot that is designed to help students brush up on their mother tongue.

The virtual bot takes the form of a ferret, which can converse with students in Mandarin, Malay and Tamil. If a student trips up – for instance, by using an English phrase in a sentence – the bot will jump in to provide the translation.

Singaporeans’ low confidence in mother tongue languages motivated the creation of this bot, said Ms Fiona Chen, a member of the six-strong team that developed it.

“Teachers also have a heavy workload, and they can only spend 10 minutes of individual time with each student in class,” said Ms Chen, a product manager who was deployed from GovTech to work in the Education Ministry’s digital excellence and products division.

LangBuddy is currently being trialled by around 300 students from 10 secondary schools and junior colleges, and there are plans to roll it out to more schools in 2026, she added.

The large-language model used is trained on the textbooks and learning materials used in the national syllabus.

Based on conversations, the bot can also give assessments of a student’s fluency and vocabulary level, and provide teachers with aggregated data of the strengths and weaknesses seen within the class.

The challenge is to continuously fine-tune the bot to make it more localised, such as training it to recognise local terms instead of those used in other countries, said Ms Chen.

“The bot is not trying to replace face-to-face conversations, but it supplements practice by giving users the opportunity to speak with something that is slightly better in the language, yet also understanding and patient enough.”