The built environment stands at a pivotal inflection point, where the convergence of Artificial Intelligence (AI) and Digital Twin (DT) technologies is fundamentally redefining how buildings are conceived, constructed, and operated. This paper presents a deeply elaborated examination of the theoretical foundations, current research landscape, and methodological frameworks underpinning AI-DT integration in building design and performance management. The global building sector currently accounts for approximately 37% of total CO₂ emissions and 30% of final global energy consumption, placing enormous pressure on architects, engineers, and facility managers to adopt transformative technologies. This study synthesises evidence from over 117 peer-reviewed studies published between 2020 and 2025, drawing on real-world implementations including IKEA's AI-driven HVAC digital twin across 42 million square feet, the Hong Kong–Zhuhai–Macau Bridge structural health monitoring system, and EU-wide smart building retrofit programmes. Key findings establish that AI-enhanced Digital Twins enable energy savings of up to 30% in HVAC operations, improve fault detection accuracy by 15–20%, and support predictive maintenance frameworks that reduce unplanned downtime by 40%. The paper introduces a five-layer AI-DT Integration Framework and a multi-phase systematic methodology for applying these technologies across the full building lifecycle. Findings reveal that while the DT market in the construction sector is projected to grow from USD 41.98 billion in 2024 to USD 93.5 billion by 2029 at a CAGR of 17.3%, significant barriers remain including data interoperability, scalability, and the energy cost of AI computation. This paper contributes a replicable research architecture and a call for standardised protocols to bridge the gap between theoretical potential and large-scale industry adoption.