AI Case Studies in Practice: When Secure Communication Breaks. Resolving a Critical Java Issue

Explore More AI Use Cases

Each scenario reflects real-world pressure points where applied AI delivers measurable impact across engineering workflows.

Case 1: Handling a Late Architecture Change Before Release
Case 2: When Secure Communication Breaks. Resolving a Critical Java Issue
Case 3: Delivering a Complex Software Analysis Under Severe Time Constraints
Case 4: Understanding and Optimizing a Legacy System Without Documentation

‍

Business Context

In enterprise environments, issues like this rarely remain isolated technical problems.

When an integration stops working, entire business processes can be affected – data exchange stops, services become unavailable, and project timelines are at risk.

Troubleshooting complex issues is typically handled by senior engineers and can take weeks, making it both expensive and risky from a delivery perspective.

Being able to resolve this type of issue in days instead of weeks means:

• lower project risk
• lower troubleshooting costs
• faster system stabilization
• fewer delays in dependent projects and integrations

This is where faster problem analysis has a direct business impact.

Situation

After a system upgrade, a Java application was no longer able to connect to a service over HTTPS. The issue occurred only on Java 11+ running on Red Hat Linux.

Why it was hard

The problem was caused by a TLS layer issue related to the random number generator (RNG) on Linux when using Java 11+. Standard troubleshooting methods failed, and generic AI chat without proper context produced misleading results. The typical approach would involve forum research, code decompilation, isolating the issue and testing multiple hypotheses.

• Estimated standard effort: 3–12 MD

How we used AI

• AI was used only after preparing structured and detailed context:
• TLS debug logs and application logs
• Detailed environment description
• Java and application version details
• List of all performed tests and results
• Decompiled test application code
• Structured description of the problem and its context
• With this level of context, AI was able to identify the RNG issue and recommend a specific configuration change ICMRMSSLRNGALGORITHM=NativePRNG

Outcome

• Standard approach: 3–12 MD
• With AI-assisted engineering: resolved in days instead of weeks
• Root cause identified and issue resolved

Delivery Impact – What this means for software delivery

• Issues like this often block integrations and can delay entire projects.
• Instead of spending weeks on troubleshooting, the issue was resolved in days.
• This reduces delivery risk, lowers troubleshooting costs, and shortens the time needed to restore system functionality.

What made the difference

• Structured context
• Engineering experience
• Iterative approach
• Validation and testing
• Domain knowledge

AI was not the solution. AI was a multiplier of an experienced engineer’s capability.