Systems | Development | Analytics | API | Testing

AI FinOps must not be an afterthought but should be a critical and mandatory component for any enterprise AI program.. The explosive growth of AI workloads from simple AI-assisted coding to product autonomous multiagent systems complicated by highly dynamic usage-based pricing has introduced next level risk in tracking and managing spend. As the Gartner report points out, 85% of enterprises have no automated controls to stop cost spikes.

Kong Insomnia 13.4 is all about giving you more visibility than ever before. Whether it's your git changes, your secrets, or how your users are using Insomnia, you’re able to see more than ever before and make better decisions than ever. We’ve gotten great feedback on the native git support we shipped in v13 a few months ago, and many teams have switched to a CLI way of working. But other teams love the Insomnia Git UI.

Today ServiceNow launched the **ServiceNow Control Tower Alliance**, and Kong is proud to be a founding member to help enterprises put AI to work safely, at scale. It's a milestone for Kong, and we think it's an important moment for the industry. Here's what the alliance is, why it matters, and why we believe every AI control tower needs a runtime. Enterprise AI has moved from pilots to production faster than most governance programs could keep up.

API regression testing re-runs your API tests after every code change so existing endpoints keep returning what clients expect. A solid setup checks response shapes, auth, and latency, mocks external dependencies, and runs a fast suite on each pull request with the full suite nightly. Keploy, an open-source tool, generates that suite and its dependency mocks from real API traffic, so teams get regression coverage without writing scripts.

Here is a situation most backend teams have lived through. A developer changes how the user service formats dates. It is a tiny change: one line, a cleaner ISO format, every unit test still green. The pull request gets approved in five minutes. Two days later, the billing service starts failing to generate invoices, because it was still expecting the old date format from the user service. Two kinds of testing could have caught that, and they are the subject of this post.