As buildings, campuses, and cities become increasingly instrumented, the challenge is no longer collecting data. The challenge is making sense of it in real time, close to where it is generated, and turning it into coordinated, intelligent action.
At Akila, this challenge sits at the heart of our platform. Our digital twin is not just a visual layer, but a living system that understands how assets, people, energy, and environments interact. To push this capability further, we are strengthening Akila’s agentic AI and orchestration engine with NVIDIA DGX Spark, bringing high-performance GPU computing to the edge and reinforcing the “brain” that powers cognition for smart spaces.
What is NVIDIA DGX Spark?
NVIDIA DGX Spark is a compact, high-performance GPU system designed to support the full lifecycle of modern AI workloads, from development and testing to deployment. It delivers the kind of accelerated computing traditionally associated with data centers, but in a form factor that can be deployed closer to operational environments.
For Akila, this means having a dedicated GPU compute foundation to run AI-intensive workloads at low latency, including:
-
Real-time video analytics
-
Sensor and meter data processing
-
AI agent reasoning and coordination
-
Spatial and simulation-driven intelligence
Rather than sending all data back to centralized cloud environments, DGX Spark allows critical AI inference and decision-making to happen closer to buildings, campuses, and districts themselves.
Why this matters for Akila
Akila is designed as an AI-native digital twin platform. We integrate real-time and historical data from equipment, energy systems, IoT sensors, and video streams into a unified, spatially anchored data referential. This foundation enables AI models not only to analyze data, but to understand context, relationships, and physical constraints.
DGX Spark strengthens this foundation by giving us the compute headroom to:
-
Develop and test GPU-heavy AI models faster
-
Run advanced inference pipelines with minimal latency
-
Prototype new AI-driven use cases without architectural compromises
In practical terms, it allows Akila to move beyond reactive analytics toward continuous reasoning and orchestration across systems.
Accelerating key capability areas
With DGX Spark under the hood, we are accelerating development across several core domains.
Agentic AI and reasoning at the edge
Akila’s agentic AI framework is designed to coordinate specialized AI agents that monitor, reason, and act across different domains, such as energy, comfort, safety, and operations. GPU-accelerated compute enables more advanced reasoning models to run closer to assets, supporting faster decisions and more autonomous behavior in complex environments.
Computer vision and spatial intelligence
Video is one of the richest but most compute-intensive data sources in the built environment. DGX Spark enables Akila to integrate computer vision capabilities such as occupancy awareness, safety monitoring, and spatial behavior analysis directly into the digital twin. These insights become spatially contextualized, allowing users to understand not just what is happening, but where and why.
Predictive building and infrastructure optimization
From energy consumption and cooling performance to equipment behavior, many optimization challenges depend on predictive, simulation-driven models. GPU acceleration allows Akila to advance these models, improving accuracy and enabling more frequent recalculation as conditions change in real time.
Akila as the orchestrator of the built environment
Beyond individual features, this step reinforces Akila’s broader role as an orchestrator and nervous system for the built environment.
Buildings and cities today rely on fragmented technologies, each optimized for a narrow function and rarely designed to work together. Akila’s platform connects these systems into a coherent digital twin, where data, AI models, and operational workflows are aligned spatially and temporally.
By pairing this orchestration layer with advanced GPU infrastructure, Akila can:
-
Coordinate multiple AI models and vendors within a single operational framework
-
Enable cross-domain intelligence across energy, operations, safety, and sustainability
-
Scale from individual buildings to portfolios, campuses, and urban districts
Built with the NVIDIA ecosystem
This work is developed in collaboration with partners across the NVIDIA Partner Network, building on proven experience deploying AI-driven digital twins for real-world environments. DGX Spark expands what we can prototype, demonstrate, and ultimately deliver, from smart buildings to smart campuses and cities.
As these capabilities are put to work in live projects, we will share more on how agentic AI, computer vision, and GPU-accelerated digital twins are reshaping how the built environment is designed, operated, and decarbonized.
More to come.
