|
|
|
|
Back to newsletter 305 contents
The Java performance industry has moved in two ways to support AI. Generally these two are firstly to provide contextual information to the AI from the outcome of a tool's investigation, and secondly to provide information to the AI on how to use a tool.
In the first approach, tools are moving to include AI models directly embedded in, giving the AI the current contextual knowledge of the issues under investigation (often using RAG). I detailed this in the "The current generation - Generation 3: Search AI powered observability" of my article covering the The past, present and future of observability. A similar approach is detailed in the second news item below, where intelligent JVM monitoring is obtained by streaming aggregated JFR events to an AI. These approaches let you ask questions directly to the AI, benefitting from the AI model's deeper understanding of the issue from the contextual information.
The second approach is by including APIs and MCP servers that give the AI information on how to use the tool. By pointing your model at the MCP server, the AI can find out what tools it can use, then it can directly use those tools to investigate issues. You can see this second technique described in the link in this month's news section about JProfiler 16.1 which includes an MCP server that let's AI Agents directly profile apps.
Now on to all the usual newsletter list of links, tips, tools, news and talks, and as usual I've extracted all the tips into this month's tips page.
Java performance tuning related news
Java performance tuning related tools
Back to newsletter 305 contents
