Writing code is getting cheaper, but reliable systems take more than code. As agents take on more of the work, the bottleneck moves to specs, tests, and review, and agents amplify whatever you feed them. Here’s what’s working across teams, what breaks, and where human judgment now pays off.

If AI can generate code and tests in seconds, do we still need QA? Yes, but the job is changing fast. Quality is stretching left into specs and acceptance criteria, and right into integration, release confidence, and production signals. QA is no longer a stage at the end of the pipeline; it’s a thread running through the whole SDLC.

TechEmpower Framework Benchmarks Round 23 surfaced something unexpected: the Spring + MongoDB implementations were performing far worse than anyone anticipated. What initially looked like a benchmarking anomaly turned into a collaborative investigation with MongoDB’s performance engineering team. This post walks through what we found, how we tracked the issue down, and what it reveals about real-world framework performance and database interaction under load.

Product specs now have two audiences: people and AI agents. If your PMs aren’t using codebase-aware tools before handoff, you’re leaving speed on the table. Here’s what we’re seeing work.

If you’re shipping LLM or RAG features, traditional QA isn’t enough. Adversarial testing, including prompt injection, data poisoning, and retrieval manipulation, is now table stakes for resilient AI.

When AI agents can act, reliability becomes architecture. A practical guide to building agentic systems that don’t run away, break trust, or rely on prompt magic.

AI coding tools promise major productivity gains, but measuring their true impact is harder than it looks. Join TechEmpower’s Tony Karrer, a DORA researcher and two CTOs for insights on the metrics that matter. Jan 9, 8–10 AM PST

Everyone’s asking whether AI tools are delivering “30% productivity gains.” They can, but only after adoption matures. Early wins hide in debugging, comprehension, and better requirements, not in lines of code.

AI is reshaping how software gets built. Join TechEmpower and O’Reilly author Brent Laster for a practical session on integrating AI tools across your SDLC—from planning to deployment.

Discover the key to successful AI coding assistant implementation with a 2-week ramp-up spike. Learn about focused training, hands-on ticket work, and feedback loops that accelerate proficiency and build shared practices.

News articles, insights, and industry updates that we find valuable.