Machines That Partner: Case Studies from Leading Malaysian Enterprises

TLDR;

• The emphasis is moving from demonstrating concepts to delivering measurable business value.

• Effective use of Agentic AI relies on human expertise, judgement, and oversight to guide its outputs.

• Organisations must design collaborative systems that integrate automation with human creativity, context, and decision-making.

• AI co-pilots enable enterprises to create highly personalised messaging tailored to diverse stakeholders at scale.

• Malaysian enterprises face challenges in AI literacy and need professionals who can connect strategy, data, and technology seamlessly.

Over the past few years, enterprises across Southeast Asia have dived headlong into artificial intelligence (AI), experimenting, prototyping, and testing its limits. It has been a playground for innovation, where “trying things out” was both the brief and the benchmark. Leadership teams were generous, offering what felt like blank checks in the name of innovation.

As organisations mature, the conversation has shifted from proof of concept to proof of value. Conversations are turning toward Agentic AI, systems designed to take initiative, collaborate with humans, and act with contextual understanding. This marks a new chapter in AI’s evolution: one that moves toward participation and partnership. At the centre of this shift is the growing recognition of the human in the loop. Done well, this balance transforms AI from a costly experiment into a genuine source of competitive advantage.

AIBP together with Adobe convened a private workshop in Kuala Lumpur, bringing together Malaysian enterprises from both from technology and marketing to co-pilot an Agentic AI project. Malaysian Enterprises at different stages of maturity worked side by side, testing how AI can complement human decision-making and extend operational capabilities.

When Power Meets Prediction: Using AI to Catch Energy Thieves

Operational inefficiencies, unclear processes, and fragmented approvals often leave potential value unrealised. For Malaysia’s energy utility, Tenaga Nasional Berhad (TNB), electricity theft represents a hidden yet costly challenge, now amplified by modern threats like illicit bitcoin mining.

“The biggest gap is not just technical know-how but the ability to connect what we’re doing today with where AI can take us. We need people who can bridge that understanding,” shared Sharman Abdul Ghani, Head ICT Intelligent Operations, Tenaga Nasional Berhad.

The focus on efficiency is not unique to TNB. According to the AIBP 25-26 ASEAN Enterprise Innovation Market Overview, 70.5% of organisations cite cost optimisation and operational efficiency as top priorities when investing in digital transformation. For utilities, manufacturing, and service companies alike, AI is increasingly valued as a lever to improve performance and reduce risk.

Historically, detection relied on tip-offs, unusual equipment movements, and unexplained spikes in night-time electricity use. The group took on TNB’s challenge and explored how Agentic AI could combine multiple signals: heat signatures from rooftops, abnormal data consumption, and surges in regional electricity draw to create a predictive picture of risk.

“Satellite data can detect heat signatures from premises. Telcos can show where data consumption is highest. And there are tools that track device behaviour patterns,” Sharman explained.

By integrating AI into these operational processes, the utility not only predicts risk but also frees up human teams to focus on validation and decision-making.

Personalisation at the Speed of Relevance

Whether in enterprise sales or consumer marketing, the goal is to reach the right person, with the right message, at precisely the right time. For Malaysian enterprises, this has been challenging: siloed systems, fragmented data, and inconsistent messaging make personalised, timely engagement difficult at scale.

Enterprises agreed that buying decisions now involve sprawling groups of stakeholders, each with their own lens and priorities. At the workshop, participants explored how Agentic AI could act as a co-pilot, helping teams craft hyper-personalised messages for each audience: the same product pitched in business terms to a CEO, in technical language to an engineer, and in pricing context to procurement.

By connecting insights from account managers, product teams, and sales, this approach maintains one continuous narrative, ensuring that AI-generated pitches and human follow-ups feel like the same conversation. On the consumer side, timing and context are critical. The Team from AirAsia, Malaysia’s leading low-cost airline group explored how an Agentic AI co-pilot could respond to cultural moments in real time.

“When an influencer posts about Lombok’s beaches, for example, the system can instantly generate a travel video promoting flights there — faster than a human can react. The content team still validates it, but AI handles the heavy lifting, allowing the brand to publish within hours and stay ahead of the social curve,” said Mervin Buid, Manager, Global Markets and HR Business Partner, AirAsia.

Using Agentic AI to craft context-aware, timely communications allows organisations to maintain consistent messaging while reaching stakeholders and customers more effectively.

The potential of AI also underscores the importance of internal readiness and careful content oversight. Gaurav Kumar, Principal Solution Consultant from Adobe recalled testing a prompt for an image of a child at the zoo, which generated a child standing next to a tiger in a cage.

“These agents look powerful, but you still need the internal expertise to use them effectively,” he said, showing that even simple outputs require human judgment and review.

Talent & Experience: Empowering People Through Intelligent Support

A common refrain from enterprises across the board was the lack of AI literacy, both among those expected to build systems and those meant to use them. Enterprise teams often lack “translators”, professionals capable of bridging strategy, data, and technology. This gap slows decision-making and adds uncertainty to AI adoption.

“One of the biggest challenges we’ve discussed in our team is the adoption of AI across the organisation. It primarily comes down to the lack of AI literacy. Some talents are supposed to be building AI, others are supposed to be using it, but very few understand both.”, said Eva Piramila Poovan, Head Group Data from Astro.

This human dimension is also apparent in education. A private Malaysian university highlighted that postgraduate students often rely on self-discipline rather than external motivation, making retention and engagement critical. Here, Agentic AI serves as a co-pilot: automating touchpoints, surfacing relevant content, and scaling communication while preserving a personalised, human-centred experience.

“We’re exploring how AI can identify what’s relevant to each student: what’s happening in their industry, how often to reach out, and through which channel. It’s about creating a learning experience that feels guided, not generic,” Puteri Sofia, Acting Head of Innovation and Learning Experience, from Taylor's University shared.

This means using agentic AI to personalise communication, surface relevant industry content, and automate touchpoints throughout the student journey without losing the human element.

Agentic AI in this context does the repetitive work, freeing humans to focus on judgement, mentorship, and interactions that require subtlety and discernment.

The Next Act

Southeast Asia’s AI journey is entering a second phase. Experimentation has given way to integration. Organisations are asking now how humans and machines can collaborate to produce insight, value, and impact.

The next competitive advantage lies in augmented intelligence, where human judgement and machine capability intersect, yielding outcomes that neither could achieve alone. The challenge for enterprises is to design systems that are cooperative, balancing automation with the judgement, context, and creativity that only humans provide.

For more conversations like this, keep an eye on our upcoming activities here.