Feature: AI Java: the strongest enabler for F


adopting agentic AI By Gaël Blondelle, Chief Membership Officer, Eclipse Foundation However, achieving that vision requires


ew trends in AI have captured developer attention as quickly as “agentic AI”, which enables intelligent soſtware agents to reason, act and collaborate autonomously across


complex IT environments; see Figure 1. But as promising as these systems are, they also pose a practical challenge for every CIO and CTO: how to deploy them within existing technology landscapes without starting from scratch. For many global enterprises that landscape runs on Java. From banking and insurance to retail, logistics and telecom, Java continues to power the mission-critical backend services that process billions of transactions every day. Traditional AI applications have mainly


been narrow and static, focused on perception, classification, or prediction. By contrast, agentic AI systems are goal- directed and adaptive. Tey don’t just answer questions; they execute multi-step workflows, make decisions and coordinate with other agents or systems, in real time. In an enterprise context, this means agents can monitor operational telemetry, correlate events across distributed systems, identify the root cause of incidents, and even remediate issues autonomously, leading to potentially immense productivity gains.


more than clever prompt engineering or connecting APIs to a large language model (LLM). It demands deep, programmatic integration with the core logic and data flows of enterprise applications. And in most enterprises, those logic and dataflows live in Java.


Java: The (not so) hidden foundation Despite being over 30 years old, Java remains the lingua franca of enterprise computing. It consistently ranks among the top languages in developer rankings, proving its enduring relevance for back- end services, financial systems, cloud- native development and other strategic business applications. Nearly 70% of organisations report that at least half of their applications run on the Java Virtual Machine. Te reasons are plain to see, as Java delivers three things that agentic AI also needs: •Maturity and stability – Decades of testing, security hardening and standardisation have made Java-based systems exceptionally reliable.


• Scaleability and performance – Frameworks like Akka, Vert.x and Quarkus enable massive concurrency and distributed state management at low latency.


Figure 1: ‘Internet of Agents’ is a decentralised network where AI agents operate autonomously


• Ecosystem and portability – Te JVM abstracts away hardware and OS differences, allowing soſtware to scale across heterogeneous environments. Enterprises have invested millions of


developer hours and decades of domain logic into Java classes, APIs and data models. Rebuilding that functionality in a new AI-native stack would be both risky and wasteful. Te smarter path is to make agentic AI interoperate with and extend those existing Java assets.


Connecting agents For agentic AI to become truly useful in enterprise settings, it must operate within existing operational, reliability and performance constraints. Agents that can’t talk to Java systems are effectively cut off from the most critical operational data, workflows and decision-making contexts. Te integration challenge comes down to


three layers: •Data access: Agents need secure access to the structured and unstructured data produced by Java applications, whether in databases, message queues, or telemetry streams.


• Process invocation: Agents must be able to call existing business logic by invoking APIs, microservices and transaction workflows implemented in Java, without introducing latency or reliability risks.


• Context sharing: Agents must understand the domain context encoded in Java models and schemas, so that LLM reasoning aligns with real-world business semantics. Bridging these layers requires


bidirectional interoperability between agentic runtimes and Java systems. Tat means agents should be able to invoke Java methods as actions, subscribe to Java events as triggers and exchange structured context without requiring manual translation.


Avoiding the “rewrite trap” Eager to deploy generative AI quickly, some enterprises have attempted to rebuild parts


30 February 2026 www.electronicsworld.co.uk


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46  |  Page 47  |  Page 48