AI systems are moving into vehicles, intersections and factory floors, where decisions often need to happen locally and quickly. As these systems become more autonomous, organizations face a practical challenge: keeping their communications secure without slowing them down.
A new solution brief from Intel and Atsign explores how secure agent communications and hardware-backed security can work together to address that challenge. Here’s what it means—and who should care.
Edge AI means running AI close to where data is created and decisions need to happen—at an intersection, inside a vehicle, on a factory floor or in a retail location.
Instead of sending every piece of information to a distant cloud and waiting for a response, an edge system can interpret what is happening locally and respond almost immediately. That makes Edge AI valuable anywhere speed, reliability or limited connectivity matters.
Edge AI can analyze information locally—for example, detecting congestion from a camera feed. Agentic AI goes further: an AI agent can interpret information, communicate with other agents and systems, make decisions and initiate actions within defined permissions.
Consider a smart transportation system. A route-planning agent might receive information from traffic feeds, roadside systems, weather services and fleet vehicles. It can evaluate road conditions and recommend a better route as conditions change.
That can improve traffic flow and help transportation operators respond more quickly. It also raises an important question: How do you know that every agent and data source is legitimate, authorized and operating within its intended boundaries?
Edge environments are highly distributed. Devices and applications may operate across cellular networks, public infrastructure, customer networks and remote locations. They cannot always depend on a traditional corporate network perimeter for protection.
Each agent therefore needs a verifiable identity. It should be allowed to communicate only with approved systems, access only the information it needs and perform only the actions it has been authorized to take.
Communications also need to remain encrypted from one authorized participant to another. Cryptographic keys must be protected, and organizations need a way to understand and audit what their agents are doing.
This is the practical meaning of zero trust at the edge: an agent or device is not trusted simply because it is connected to the right network. Identity and authority must be established before communication begins.
Encryption requires computing resources. On edge devices, it may compete with AI and other time-sensitive tasks for limited computing resources. If securing communications slows the system’s response, it can undermine the reason for processing information locally in the first place.
A traffic-management system cannot wait indefinitely to respond to an accident or sudden congestion. The same is true for factory equipment, energy infrastructure and other systems where digital decisions affect physical operations.
Organizations should not have to choose between strong security and timely decisions. The challenge is to make encrypted communication practical at the speed and scale these systems require.
Intel and Atsign address different layers of this challenge.
Atsign provides the security architecture for communication between agents, applications and devices. Each participant receives its own cryptographic identity and can access only the information explicitly shared with it. End-to-end encryption protects communications between authorized parties. The architecture also supports auditability and the ability to change or revoke authority as requirements evolve.
Intel provides the underlying compute and hardware security capabilities. These include hardware-backed key generation, accelerated encryption and hashing, protected key storage, secure boot and platform-integrity protections.
Atsign determines who and what may communicate, while Intel’s hardware helps protect and accelerate the cryptographic operations supporting those communications.
Testing conducted by Atsign measured encryption throughput with software-only encryption and with Intel hardware acceleration:
The relevance to the joint solution is straightforward: Atsign provides the architecture for secure agent communications, while Intel hardware acceleration helps make encryption practical at scale.
These are encryption-throughput measurements, not overall application-speed measurements. An approximately 88x improvement does not mean an entire AI system runs 88 times faster. Actual performance varies by use, configuration and other factors.
This is especially relevant to organizations building or operating systems in which AI can interact with real-world infrastructure, including:
Although the joint solution brief focuses on smart transportation, the underlying challenge is much broader. Any organization deploying AI agents across distributed devices, applications and networks will need to answer the same questions about identity, authority, communication and performance.
The goal is not merely to make AI available at the edge. It is to make sure that autonomous systems can communicate and act without exposing the infrastructure and data on which they depend.
Read the Intel and Atsign solution brief: Securing Agentic AI at the Edge for Smart Transportation
Exploring an Edge AI deployment? Talk with Atsign about securing communications between the agents, devices and systems involved.
Testing conducted by Atsign as of June 17, 2026. Performance varies by use, configuration and other factors. See the solution brief for complete test configurations and disclaimers.
