Operations | Monitoring | ITSM | DevOps | Cloud

Uber blew its annual AI budget in 4 months

Uber burned through its entire annual AI budget in under 4 months. Here's what went wrong — and what every engineering org should be doing instead. The data: 80% more code is getting pushed with AI… but only 18% of AI-written code actually ships to production. That's not a productivity story. That's a spend problem. If you're scaling AI tooling without real-time monitoring and guardrails, you're Uber.

AI Might Break Open Source Differently Than You Think

AI coding agents may not replace open source libraries overnight. But Adam Arellano, Field CTO at Harness, thinks models like Mythos could expose a bigger problem: finding bugs, vulnerabilities, and edge cases faster than maintainers can keep up. That might be the real threat to tools and libraries.

AI, Platforms, and the Future of Value Delivery: A Conversation with ServiceNow

How do enterprises turn AI from experimental potential into real-world software delivery value — without slowing down, breaking security, or sacrificing reliability? At {unscripted} 2025, Amit Zavery — President, Chief Product Officer, and COO of ServiceNow — joined Harness CEO and Founder Jyoti Bansal for a candid fireside chat on the future of AI in the enterprise, the role of platforms in unlocking developer productivity, and why"AI-native" only works when speed, security, and reliability move together.

AI Asked Our General Counsel Anything. She Didn't Hold Back.

What happens when AI interviews a tech leader? You get unexpectedly honest answers. Harness General Counsel Hanna Steinbach sat down with ChatGPT — and skipped the corporate script. From the realities of parenting while leading a legal team at a high-growth startup, to the daily habits that keep her grounded, this is the kind of candid leadership perspective you rarely see. Oh, and she's definitely the person sprinting to the gate right as boarding starts.

SLI, SLO, SLA: What They Mean for Load Testing

Most engineers can recite these three terms. Fewer know how they actually connect during a load test. If your team is running performance tests without mapping results to SLOs, you're collecting data without a pass/fail signal. This short gives you the mental model to turn load test output into something your SLA can actually depend on.