AI GUIDE FOR BOARDS IN SINGAPORE
Section V: Organisational Resilience and Capability
9.5 Embedding AI
AI creates more value when integrated into existing systems, workows tools employees already use, rather than operating in isolation. AI systems that remain disconnected from core business platforms o en deliver limited operational impact. For example, an AI- enabled reconciliation capability provides little enterprise value if it cannot integrate securely with nance, enterprise resource planning, or workow systems. Boards should therefore ensure: • Integration discipline: AI systems connect securely to enterprise platforms (such as Enterprise Resource Planning, Customer Relationship Management and Human Resources), collaboration, analytics and operational systems through governed application programming interface and enterprise architecture standards. • Interoperability standards: Adoption of open standards and interoperable architectures (e.g., Model Context Protocol) reduces technical debt, minimises vendor lock-in, and supports long-term resilience. • Workow alignment: AI capabilities are embedded within existing approval chains, governance controls, audit trails, escalation pathways and human oversight mechanisms.
to interact securely with document repositories, drive storage, collaboration tools, communication systems (e.g., email, calendars), operational platforms, codebases, analytics tools and internal systems. is makes AI more context-aware, more actionable, and more useful in day-to-day operations. Company knowledge and application integration are closely linked. AI systems become more useful when they can access trusted organisational context across connected systems and generate outputs that are speci c to the organisation’s business, rather than generic in nature. Boards should therefore look beyond whether AI tools exist in the organisation and ask whether they are interoperable with the workows, data sources and operational systems where enterprise value is actually created. At scale, seamless integration is not just a user-experience bene t; it is part of resilience. AI systems that are better connected to real workows are more likely to be adopted, governed, monitored and relied upon as part of system operations.
Boards governing multi-vendor AI estates must insist on architectural coherence to avoid data silos and duplicated governance overhead.
9.5.1 Avoiding data silos
As AI becomes more embedded in work, value increasingly depends on its ability
88
Made with FlippingBook Digital Proposal Creator