From its national artificial intelligence (AI) roadmap to its smart city ambitions and growing startup ecosystem, Malaysia is stepping confidently into the age of AI. Businesses, especially small and medium-sized enterprises (SMEs), are experimenting with chatbots, predictive analytics, and AI-powered customer tools, but the next stage requires more than just experimentation.
AI has now reached a stage where it can accelerate enterprise productivity. To truly transform productivity and unlock value at scale, Malaysian businesses need to shift their focus from chasing the biggest or newest models to implementing AI that deeply understands their business, data, and goals. This means leveraging AI to reliably automate mundane and time-consuming processes, ultimately freeing up your team to focus on more meaningful and innovative work.
Bridging the gap between possibility and authenticity will define 2026. This is the year that companies stop chasing bigger models and start demanding smarter, contextual models that fit their needs.
There are six important movements. Build agents that reason on your own data, work in coordinated teams, are continuously evaluated, work across modalities, seamlessly integrate into your workflow, and are supported by people trained to work with you.
#1 Migration from general-purpose AI agents to domain-specific AI agents
Generic models trained on public internet data still struggle with the messy realities of enterprise processes because they lack deep organizational context. Additionally, in today’s regulatory and geopolitical climate, businesses face increasing demands for data and AI sovereignty, ensuring data privacy, security, and compliance within a given jurisdiction or business environment.
Domain-specific agents are based on proprietary data with controlled lineage and more accurately interpret internal rules, edge cases, and compliance constraints, as well as maintain important sovereignty requirements. This control over data and AI models reduces risk, meets legal and ethical obligations, and maintains competitive advantage.
We have already seen this play out with Digital Nasional Bhd (DNB), which is spearheading Malaysia’s 5G rollout. Facing massive amounts of 5G data and the challenge of providing real-time insights across a growing network, DNB leveraged the Databricks Data Intelligence Platform to build a secure, cost-effective, and high-performance AI and analytics foundation. The result is a 70% increase in data pipeline performance and cost optimization, while enabling smarter network planning and operational decisions.
#2 Migration from single agent to multi-agent orchestration
Business work is rarely done in one step, and enterprise AI is no different. Real-world workflows span search, validation, approval, and decision-making across multiple systems and teams, far beyond what a single agent can reliably handle. The next phase is multi-agent orchestration. Here, specialized agents handle tasks such as compliance checks, data retrieval, and inference, and supervisory agents coordinate these tasks.
Supervising agents enable organizations to scale AI beyond isolated pilots into managed, auditable, and adaptable workflows by ordering roles, delegating work, and synthesizing results in natural language.
#3 Moving from one-time checks to continuous evaluations
Once AI goes into production, continuous real-time evaluation becomes non-negotiable. Models that appear powerful in training often degrade on real data, drift as inputs change, and without continuous evaluation, their reliability quickly declines. Next year will see companies adopt assessment-centric practices, where agents are continually evaluated against real-world tasks, real-world feedback, and changing conditions.
Agent Bricks is built on this principle. This means streamlining the development of domain-specific agents, enabling teams to define objectives and quality metrics in natural language, automatically generating test suites, and optimizing performance based on enterprise data. By creating an environment where AI evaluates AI, companies can reduce uncertainty, accelerate adoption, and enable agents to continuously learn from successes and failures to better match specific needs.
#4 Transition from text to multimodality
While AI has traditionally been text-first, both consumers and businesses now communicate using a combination of voice notes, videos, screenshots, sensor feeds, and chat messages. Multimodal AI fits into this reality by understanding and combining these diverse inputs, dramatically expanding the scope of what can be automated in real-world operations.
In practice, multimodal workflows greatly enhance human interpretation. Customer service AI agents can read your messages, analyze your tone of voice, and interpret screenshots and videos of issues. In healthcare, models can fuse patient records, medical images, and sensor data to support more accurate diagnosis and personalized treatment plans. In retail and e-commerce, multimodal agents can process reviews, product images, and usage videos to better understand customer preferences, improve recommendations, and identify fraud.
#5 Moving from AI as a feature to invisible integration
The most successful AI systems do not advertise themselves. They blend into your workflow and silently increase productivity without causing friction for your employees or customers. Invisible AI means automation is built-in, consistent, and intuitive. This becomes an environment within which the team operates, rather than a feature that the team must learn how to use. When systems are continuously evaluated, humans and AI can seamlessly work together to accelerate work.
#6 Continued focus on skills
As AI agents become embedded in daily operations, organizations must continue to invest in their talent. This includes not only building systems, but also teaching them how to manage, coach, and collaborate with them. You don’t have to be a data expert to reap the benefits. For example, marketers who automate data entry primarily need prompting and workflow skills to instruct AI agents to take over the work.
2026: AI will help you understand your business better
These six changes will redefine how companies use AI. Domain-specific sovereign agents make AI business-ready. Orchestration, continuous evaluation, and multimodality make it reliable and effective at scale. Invisible integration and strong skills make it a natural fit for your daily work.
The organizations that will win in 2026 will not be those with the largest models, but those that build strong data and AI governance, deploy domain-aware agents as trusted colleagues, and continue to evolve the way their employees and systems learn from each other.
Nick Eayrs is Vice President of Field Engineering (Asia Pacific and Japan) at Databricks.
