Intel, Atsign boost secure agentic AI for transport

Intel and Atsign have introduced a security and performance solution for agentic edge AI in smart transportation systems. Their joint work improved encrypted agent-to-agent communications by about 88 times on some Intel hardware.

The collaboration addresses a core challenge in running autonomous AI systems at the edge: maintaining end-to-end encryption without sharply reducing speed. It combines Atsign's identity-based communications architecture with Intel's hardware-based security features to reduce that trade-off.

According to the companies, the benchmarks target transport networks that rely on distributed decision-making across roadside infrastructure, intersections, traffic corridors and connected mobility systems. In those environments, software agents and devices must exchange data quickly while also verifying identities, securing cryptographic keys and protecting workloads.

The work is outlined in a smart transportation solution brief focused on securing agentic AI at the edge. It describes a zero-trust model in which devices, AI agents, data exchanges and workloads are authenticated and protected across decentralised transport infrastructure.

Benchmark results

The headline performance figures vary by processor. Intel Xeon systems showed an improvement of about 88 times, with throughput of up to 5 Gbps, while Intel Core Ultra 9 systems showed an improvement of about 60 times, with throughput of up to 6 Gbps.

Those figures matter because encryption overhead has often limited the deployment of AI systems outside centralised data centres. In transport environments, that can affect use cases that require local, low-latency responses, including traffic management, roadside sensing and automated operational decisions.

Atsign's contribution includes trusted identity management, namespace isolation, scoped data sharing, end-to-end encryption and auditability. Intel's contribution includes hardware-backed key generation, encryption and hashing support, protected key storage, secure boot, trusted compute, total memory encryption and full disk encryption.

Together, those elements are intended to secure both the communications layer and the underlying machine. The model is designed to help transport operators and developers verify that participating devices and software agents are genuine before exchanging sensitive operational data.

Transport focus

Smart transportation has become an early target for edge AI because decisions often need to be made near the point of action rather than in distant cloud environments. That can include co-ordination between signals, sensors, control systems and autonomous software operating across a wider mobility network.

As those systems become more autonomous, the security challenge extends beyond network protection alone. Operators also need to manage device identities, cryptographic keys, workload integrity and access controls across large numbers of endpoints, many of which may be exposed or physically dispersed.

The solution was built to address that broader trust problem. Atsign says its platform authenticates participants before a connection is established and protects data end to end without exposed inbound ports or reliance on static IP addresses.

The tie-up also expands Atsign's position within Intel's industrial ecosystem. Intel has added Atsign to its Industrial Builders partner programme, giving customers access to Atsign through a partner directory that highlights tested solutions and technical specialisms.

The listing is intended to simplify sourcing and reduce integration risk for customers seeking products that work with Intel architectures. It also signals that Atsign meets Intel's compliance standards and can draw on technical support for more complex deployments.

For Intel, the arrangement supports its effort to position its processors and security functions within industrial and edge AI applications, where chip groups are competing to show that local AI workloads can run securely as well as efficiently. For Atsign, the partnership offers a higher-profile route into transport and industrial projects where trust and performance are often evaluated together.

The announcement comes as businesses and public-sector operators test more autonomous AI models outside traditional IT environments. In those settings, encryption remains essential, but the processing cost can become a practical barrier when systems must make rapid decisions or maintain high-volume machine-to-machine communications.

Aparna Rayasam, Chief Executive Officer of Atsign, described the development in stark terms. "Intel and Atsign, representing the forefront of edge hardware and trusted identity management respectively, have today delivered a performance breakthrough: a genuinely secure foundation for agentic edge AI deployments in smart transportation environments, which are no longer constrained by something as mundane as 'technical performance'. But make no mistake: this is a game-changing evolution for anyone involved in the development of any system in which AI-enabled devices are required to operate autonomously at the edge, where both trust and speed are essential," Rayasam said.