The integration of large language models (LLMs) with robotics represents an emerging approach for improving intelligence, automation, coordination, and sustainability in construction management. This research and review paper develops a conceptual framework for integrating language-based intelligence with robotic execution in construction environments. The analysis is grounded exclusively in the seven references supplied by the study, which primarily investigate Tibetan-language sorting, sequencing, recognition, Unicode-based processing, algorithmic ordering, and rule-based computational implementation. Although these references do not directly address construction, robotics, sustainability, or contemporary LLMs, their computational principles provide a useful theoretical basis for understanding how structured information can be transformed into machine-interpretable sequences and operational rules. The proposed framework therefore conceptualizes an LLM as an interaction and reasoning layer, a structured information layer as the translation mechanism, and robotics as the physical execution layer. Particular attention is given to information normalization, task sequencing, rule representation, human–robot coordination, resource optimization, and sustainability-oriented decision support. The findings indicate that the fundamental challenge is not merely connecting an LLM to a robot but establishing reliable transformations between ambiguous natural-language instructions and deterministic physical actions. The paper argues that rule-based sequencing and structured information processing remain important foundations for reliable intelligent construction automation. Limitations arise from the indirect relevance of the supplied literature and the absence of empirical construction datasets. Future research should therefore validate the framework through construction-specific experiments, digital twins, robotic field trials, and sustainability performance measurements.