SGH’s citizen developers drive a bottom-up approach towards innovation
By Si Ying Thian
Singapore General Hospital is empowering its non-technical staff to develop automation and AI solutions to tackle pain points in their daily work.
-1747980687028.jpg)
RPA citizen developers at Singapore General Hospital (SGH). Image: SGH's AI & Automation team
A citizen developer revolution is brewing in Singapore’s largest tertiary hospital, the Singapore General Hospital (SGH).
More than 227 employees with non-technical backgrounds are developing automation and artificial intelligence (AI) solutions for their daily work.
This bottom-up revolution is led by the AI & Automation (AIA) team under SGH’s Department of Future Health System.
For the six-person strong AIA team, adopting a citizen developer model was key not only for managing the high demand for automation solutions – but also for ensuring solutions truly hit the mark.

This “federated approach” – according to both SGH's Manager of AIA, Chan Wai Ching, and Senior Manager of AIA, Jonathan Tan – was chosen over a central IT team approach due to its feasibility and scalability.
A federated approach lets individual departments create their own models to fit their unique needs and maintain them more effectively.
The most effective solutions emerge from users themselves, when they overcome their own pain points at work, rather than a result of technology being pushed from top down, they say.
“The best way is for people who are doing the daily work to build the tool and see the possibilities to connect the dots,” says Tan to GovInsider.
To subscribe to the GovInsider bulletin, click here.
Initial focus on RPA
SGH’s citizen developer programme has since expanded from its robotic process automation (RPA) focus back in 2019, to include AI solutions like chatbots last year.
RPA was a natural inclination given the many manual processes within the operations side of the hospital, says Chan.
While the team initially relied on an external tech provider to teach SGH staff how to build automation solutions, it lacked the specific relevance needed for the healthcare setting, she adds.

This gap prompted SGH to pivot to in-house training, which proved crucial for scaling their citizen developer programme.
Currently, the AIA team still manages RPA projects for departments without citizen developers, but those with experienced citizen developers largely handle their projects independently, seeking occasional guidance, she says.
Among the experienced citizen developers interviewed by GovInsider were Shawn Poh Kiat Keong from Specialist Outpatient Clinic Operations and Siti Nurkiah Binte Mohd Amin from Division of Medicine. Both of whom possess some coding knowledge.
They found the specialised peer support from the AIA team more helpful, particularly when it came to shared scripts, as well as external resources like online videos, forums and ChatGPT.
RPA and AI are complementary
Tan highlights that RPA and generative AI (GenAI) are complementary, and there are upcoming plans to integrate both technologies.
Adopting RPA alone has its limitations as it is designed for structured data and rule-based tasks, he notes.
He adds that he witnessed instances where citizen developers integrated government AI chatbot Pair into their RPA workflows to handle unstructured tasks as part of end-to-end automation.
According to Tan, Pair was only made available to SingHealth, the healthcare cluster that SGH operates under, since June last year.
Earlier this year, his team worked with Open Government Products (OGP) to organise a hackathon to equip SGH staff with the skills to develop AI chatbots for their individual use cases.
The hackathon was open to the whole-of-hospital, and Tan notes an "encouraging" and "overwhelming response” with 150 participants across 22 clinical and non-clinical departments having created 48 AI assistants.
The teams were guided by the AIA team and OGP to build, test and refine their AI bots.
Less steep of a learning curve with AI
GenAI tools like Pair also foster widespread adoption, given their ease of use relative to RPA, which the latter requires some technical knowledge to write an automation script, says Tan.
Pharmacist Lye Jian Cheng’s team participated in the hackathon and created a centralised chatbot for pharmacy department staff to access workflow information and legal requirements.
They were inspired by another AI assistant developed using Pair, Proph Abby, that streamlines surgical antibiotic prophylaxis guidelines for doctors.
“[Pair] was simple to use, even without any coding background. It was very intuitive to use. Initially I had to play through it, tweak the instructions, upload the reference documents and see how the chatbot responds,” says Lye.

Two citizen developers from SGH’s Visitor Services, Vinay Kumar and Cheryl Heng Xue Qi, took a different approach. They just wanted to host their AI chatbot on a secure government platform like Pair.
With over 70 staff working varied shifts, disseminating consistent information is challenging as not all staff can attend briefings in person, says Kumar.
Their chatbot has made information retrieval and clarification easier. The chatbot went live last August.
Beyond immediate efficiency gains, Heng notes how employee morale has improved.
Newer employees, in particular, feel less stressed and overwhelmed by the vast amount of information they need to know, she says.
She adds that the chatbot has allowed them to quickly access details, reducing the need to remember everything and enabling them to focus on higher-value tasks.
Beyond one-hit wonder: Sustaining the use
OGP also supports SGH in creating a dashboard for the AIA team to have visibility over the bot creation and usage across the hospital.
“Assessing and managing risk that may arise from the use of GenAI tools is a priority for us, so having visibility is important.”
“This is so we can see what the use case is and if it is veering more into the clinical side that will have impact on patient safety,” says Tan.

Additionally, he adds that beyond training SGH staff to create bots, the team is focusing on training critical evaluation and testing skills for staff.
This involves not just self-testing, but rigorously validating solutions with actual end-users and defining clear metrics of success, he shares.
He adds that SingHealth’s Office of Digital Empowerment (ODE) also organises regular, hands-on Pair workshops that are open to all SingHealth staff.
Participants leave the workshops with a working bot, and this has proven effective based on feedback from past participants, he adds.
To maintain a sustained engagement, the AIA team is also putting in place an award for “Sustained Bot Usage” for the hackathon, emphasising that the goal is not just creating one-shot wonders but impactful and usable solutions.