The AI Assistant is now available in Azul Intelligence Cloud. It lets you ask a question about your Java estate the way you would think of it, in a sentence or two, and get an answer back.
Intelligence Cloud already knows what is running across your estate and which parts of your code actually get used. That part is not new. What is new is the distance between the question in your head and the answer on your screen. You think in questions, so now you can just ask one.
Ask about what Java you have
IT operations teams spend a lot of time establishing where Java is running and what version it is. With the Assistant, that is one question:
Where have I used [Vendor] Java in the last 30 days?
The answer comes back as the hostnames, with the vendor and version on each one. That is the JVM Inventory doing the work underneath, but you did not have to ask for a report. The answer is similar to a SAM tool but you can add additional parts to the question like “what is the main method” and the answer will contain what Java actually invoked.
Ask about what code you can deprecate
For companies that have built Java software for years, most software contains remnants of old features. Things that were done one way before where the old way was never removed. In Azul’s 2026 State of Java Survey & Report, 63% of respondents reported that dead or unused code is hurting DevOps productivity. It still gets compiled, scanned, reviewed, and read, and now it gets read by the AI assistants working in your codebase too. Every unused class an assistant loads to understand an area is context window it did not spend on the actual task. Teams spend time working around this code, testing it, and doing various amounts of unnecessary work.
Static analysis and IDEs already catch part of this — code that is not called anywhere. The bigger category is code that has callers, passes the linter, and still never runs. Code Inventory sees that, because it works from what actually executed in production rather than what could execute on paper. So you can ask something like:
My AppEnv is MyApp_b123_prod and my own namespace is com/example, what package in my namespace should I start with to deprecate the most unused code?
You get back a couple of packages to start with, ranked by how much unused code is sitting in them. Not a list of everything, just the place to begin. Every line that comes out is one your developers do not have to read and one your assistants do not have to load. We refer to this as “recovered capacity” because your team gets time and attention back to focus on the important things.
Good scope gets good answers
Both of those questions have something in common: they name a thing. A vendor and a time window. An AppEnv and a namespace. The more specific you are about what you are asking about, the more directly useful the answer is.
Turning it on
The AI Assistant is available to all Intelligence Cloud customers. Enabling the AI features requires a new agreement (at no cost) that covers the use of AI-assisted answers. Ask your Azul contact and we will get it switched on.
From there, the fastest way to see what it does is to open Intelligence Cloud and ask it a few questions.
Frequently Asked Questions
How can I find unused or dead code in production Java applications without relying on static analysis tools?
Static analyzers routinely flag code as “dead” that’s actually invoked through mechanisms like the popular Spring dependency injection, reflection, or other dynamic paths, so teams stop trusting the reports. Azul Intelligence Cloud’s Code Inventory solves this with runtime-based detection: it instruments the JVM itself while applications run in production with no performance impact, then lets teams query that live execution data through the Azul Intelligence Cloud AI Assistant to see which classes and methods actually executed versus which never ran. That distinction — declared versus actually invoked — is what turns technical debt identification and removal into a more defensible, more accurate, and safer motion.
How do enterprises maintain a continuously more accurate inventory of every JVM running across on-prem, cloud, and containers?
Point-in-time JVM inventories, manually maintained spreadsheets, and scheduled scans are all out of date before they’re finished, largely due to the proliferation of ephemeral Kubernetes pods and continuous integration (CI) / build farms that include JDKs. Azul Intelligence Cloud’s JVM Inventory performs continuous discovery that keeps a live catalog current as instances spin up and down, so IT and DevOps teams can use the AI Assistant to query an up-to-date inventory instead of trusting a snapshot.
Can IT, DevOps, and security teams ask questions in natural language about their Java estate instead of building custom reports?
Yes. This is exactly what the Azul Intelligence Cloud AI Assistant is built for. It’s a natural-language query interface that lets teams ask conversational questions like “which JVMs are running Java versions which are not the latest updates?” or “where is dead code accumulating?” and get answers pulled directly from JVM Inventory and Code Inventory, with no summary or report-building steps and no performance impact from the underlying data collection.
How can I tell if a Java runtime I thought I’d retired (e.g., an old Oracle JVM) has reappeared in production?
Post-migration drift is common: a rollback, a forgotten node, a shadow deployment, or various scripts and processes that haven’t been updated can reintroduce an Oracle Java runtime, exposing the enterprise to compliance and licensing risk. Azul Intelligence Cloud’s JVM Inventory continuously monitors the Java estate and detects when a supposedly retired Oracle JVM runs again, so teams can pinpoint when and where it came back into production using the AI Assistant, rather than discovering it by accident.
How can enterprises respond to a world where AI can find and exploit vulnerabilities in hours, not weeks?
Frontier models like Anthropic’s Mythos and OpenAI’s Aardvark have shown that AI can now discover previously unknown vulnerabilities and build working exploits with minimal human involvement — a preview of a capability expected to reach the broader market, including adversaries, very soon. That means security and IT teams should no longer be asking “can we patch fast enough?” but rather “can we assess, right now, whether our Java estate is up to date?” Azul Intelligence Cloud closes that gap by keeping JVM Inventory continuously current, so the moment a new security update is released, or worse, a zero-day exploit is disclosed, teams can act.
