Agentic AI transforms siloed systems into seamless operations

One team at Singapore’s Infocomm Media Development Authority (IMDA) wanted to automate its scholar management process to make it more efficient and user-friendly.

Working with Workato, an intelligent automation and integration platform, IMDA implemented low-code technologies, intelligent workflows and artificial intelligence (AI)-powered chatbots.

Within six months, IMDA developed over 50 workflows, automated processes from scholarship requests to reporting analytics, and even integrated chatbot interactions for efficient information retrieval and service delivery – saving more than 1,000 man-hours over 12 months.

This has the potential to become a cohesive and fully autonomous ecosystem powered by agentic AI.

Workato’s Director, Solutions Consulting, Asia-Pacific and Japan, Ken Ng, likened agentic AI to be a “digital twin” or a “nervous system.” Beyond its ability to define and recommend the next steps, a key enabler of agentic AI was its ability to understand the context of the problem and then execute the next steps.

Having implemented a few agentic AI solutions across the industry, Ng shared some of the most common use cases and the key enablers for public sector agencies to leverage agentic AI.

To subscribe to the GovInsider bulletin click here.

Quick wins in internal use cases

Introducing agentic AI within internal operations provides agencies with a controlled environment to assess its tangible benefits, while also mitigating potential public concerns associated with deploying AI in frontline service delivery, said Ng.

The use of agentic AI in the workplace was also mentioned by IMDA’s Assistant Chief Executive and Deputy Commissioner of PDPC, Strategic Policy & Operations, in GovInsider’s Women in GovTech 2024 report.

“Agentic AI systems have the potential to transform the workplace – we have to understand how to deploy them widely, and responsibly,” she said.

In one case, Workato partnered with a pharmaceutical firm to scope the implementation of agentic AI for employee productivity. It was as simple as to automate meeting summaries and sending of these emails to all meeting attendees – without any prompting.

This led to the adoption of both generative AI (GenAI) and agentic AI for marketing content creation.

Impressed by the results, the firm was now planning to implement agentic AI in their IT help desk to streamline operations and free up staff for complex tasks and customer service.

Data and systems integration

“It’s one thing to hear about agentic AI in theory. It sounds great, but implementing it starts to reveal where the challenges would be,” he explained, adding that these challenges range from data access and systems integration, to building a robust AI knowledge base.

From Ng’s conversations with public sector agencies, data sharing remains a pain point – be it between agencies, or even between departments in the same agency.

Agencies are prioritising the creation of robust data platforms and data consolidation. This is a key enabler to develop contextually aware large language models (LLMs).

Workato supports this process by working with agencies to transform sensitive data into potentially shareable formats to be uploaded on these common data platforms, which then fuel the AI models, he added.

“Once the foundation is there and the integration is done, that’s where a lot of the ideas will start to come, and these agencies can really then embark on the agentic AI use cases,” he noted.

For agencies who are managing a mix of legacy systems and newer solutions like Software-as-a-Service (SaaS), Ng said that they can tap on middleware solutions to connect these systems and make them “speak the same language.”

Key enablers for agentic AI

Ng highlighted three key factors for the successful implementation of agentic AI in the public sector: Clear strategy, executive sponsorship, and data integration.

To get management buy-in, AI agents need to be tied to a business problem or an objective.

“Rather than just jumping in straight, I would say have a clear strategy and understand what [problems] the technology actually addresses for the organisation,” he noted.

An executive sponsorship, which is the buy-in from multiple stakeholders across business, customer service, and IT units, would enable cross-department collaboration required to seamlessly implement agentic AI.

The foundation of security is also making sure there is a clear understanding of AI policies among the officers who are contributing the data to these platforms and tapping on AI for public service delivery, he added.

“While AI is still in a hype cycle at this point of time, I would say that agentic AI is a game-changing differentiator here – as a productivity agent.

“The key thing is understanding how to implement it and what value it gives to the organisation.

“If organisations can define that, I think that’s where they will start to quickly find success in its implementation and not have it as a white elephant science project,” he said.

Workato’s Director, Solutions Consulting, Asia-Pacific and Japan, Ken Ng, will be speaking as a panelist in an agentic AI panel discussion at GovInsider’s Festival of Innovation (FOI) 2025 in March. The session is on March 26, from 11.45am – 12.30pm. You can find out more about FOI and register here >>>