Core Insights - The article discusses the launch of a comprehensive spatial intelligence strategy by Teslian, centered around the Space-Aware Large Model (Space-Aware LM) and the Space-Aware Agent, aimed at practical applications of spatial intelligence and enhancing user experience in the era of large models [1][2]. Group 1: Spatial Intelligence Concept - Spatial intelligence refers to the unified ability to understand, reason, generate, and act within a dynamic three-dimensional world, achieved through three core elements: spatial elements, spatial models, and spatial agents [2][4]. - The spatial agent acts as the main executor in a digital-physical integrated world, capable of interacting with humans through language and behavior, and processing various spatial-related tasks [4]. Group 2: Capabilities of Spatial Agents - The spatial agent exhibits five core capabilities: spatial generation, spatial perception, spatial interaction, spatial execution, and intelligent evolution, enabling it to create a "digital parallel world" and perceive physical parameters and environmental data [4]. - It can interact with humans using natural language, accurately understand intentions, provide spatial information and analysis, and execute tasks in the physical world, creating a feedback loop between the physical and digital realms [4]. Group 3: Service Framework - Teslian has developed an "N+1+1+5" spatial intelligence service system, which includes multiple spatial elements supporting a single spatial model for decision-making, coordinated by one spatial agent to empower five types of service spaces [5]. - This framework allows for the creation of a new business model termed "Space as a Service," transforming how services are provided by enabling users to make vague natural language requests, which the spatial model can interpret and act upon [6]. Group 4: Impact on Business and Society - The shift to spatial intelligence as a service will significantly enhance execution and management efficiency, fundamentally changing the understanding of intelligent scenarios and how spaces serve human needs [6].

2024年以来，基础模型能力的不断跃迁，使得作为第三代AI核心方向的"空间智能"加速涌现。 空间智能意指在动态三维世界中理解、推理、生成和行动的统一能力 ， 在特斯联的视角中，其能力 主要通过 "空间要素"、"空间模型"、"空间智能体" 三大核心要素实现，亦即：基于对物理空间内 空间要素的感知、理解和处理，以空间大模型作为核心计算与推理引擎，使空间智能体具备自主决 策、执行以及交互的智能能力，从而建立物理世界与数字世界的无缝链接和融合，进而实现万物互 联、跨越虚实、空间觉知的AI能力。 空间智能实现的三大核心要素：空间要素、空间大模型、空间智能体 近日，特斯联正式发布完整空间智能战略，构建以空间智能大模型（Space-Aware LM）为核心驱 动的空间智能体（Space-Aware Agent），以实现空间智能落地实践，升级在大模型时代的全新空 间智能体验。 空间智能体还可以通过自然语言和人交互，精准理解人的意图，提供空间信息数据和分析，同时执行 和操作物理设备，实现物理世界的反馈，实现从物理世界到数字世界再到物理世界的任务执行闭环。 随着数据积累和行为变化，空间智能体将进行周期性的强化学习，提高空间理解的 ...