Systems | Development | Analytics | API | Testing

AI now writes 30% of Big Tech’s code, but the resulting surge in defects is crashing platforms like AWS and GitHub. Manual testing can no longer keep up with this velocity; it's time to deploy AI Quality Agents to save our systems. Is AI speed worth the decline in code quality, or are we headed for a breaking point? Let me know if you’ve noticed more bugs in your workflow lately. Video collab with @ScottMooreConsultingLLC.

The "Velocity Gap" is real. AI like Claude and GitHub Copilot are pumping out code faster than ever, but there’s a catch: Engineers don't trust it yet. We’re moving away from the old days of "clicking around" in a test environment, but how do we verify code at the speed of light? Ken breaks down why the future of QA isn't just "testing," it’s simulation. Video collab with @ScottMooreConsultingLLC Learn More: speedscale.com.

If you're spinning up a whole container just for one test, you’re doing it wrong. Old way: Full DB container + pg_restore New way: speedscale + proxymock It records actual DB traffic and mocks it "on the wire." Test smarter, not harder.

If your system is already down, you aren't being proactive; you're being reactive. True proactivity happens before the release. Validate your code before it hits production. Learn more: speedscale.com.

Stop asking AI to analyze production data without context. If you want AI to write better code, you have to feed it the right data. Learn how at speedscale.com.

Learn more: speedscale.com.

Are your LLM coding agents (like Cursor or Claude Code) hallucinating fixes because they don't know what's actually happening in production? In this video, Matt from Speedscale shows you how to bridge the gap between your local IDE and live production traffic using the Model Context Protocol (MCP). Most observability tools just give you telemetry. Speedscale’s MCP server gives your agent the "inner workings" of actual API calls and payloads, so it can check its assumptions against reality. No more "vibe-coding" and hoping it works; let your agent find the 500 errors and rate limits for you.

Learn more at speedscale.com or visit https://docs.speedscale.com/proxymock/how-it-works/mcp/

Learn more at speedscale.com or check out https://docs.speedscale.com/proxymock/how-it-works/mcp/

Thinking about learning Kubernetes in 2026? You’ll need GitOps, kubectl, and CI/CD pipelines... OR you can just use Speedscale. See how a single operator replaces a million dependencies and gives you the traffic insights you actually need to survive production.

Is Kubernetes actually hard, or are we just using the wrong tools? In 2026, the Kubernetes ecosystem has become a "dependency jungle." Between GitOps, YAML configuration, kubectl mastery, and complex CI/CD pipelines, developers are spending more time managing infrastructure than writing code. In this video, Ken breaks down the "hard parts" of K8s and introduces a more efficient workflow using Speedscale. Learn how to gain instant visibility into your cluster, pull logs without the headache, and turn real-world traffic into actionable load tests.

gRPC is fast and efficient - until it breaks at 2:00 AM. Traditional observability tools are built for HTTP/1.1 and JSON. When you switch to gRPC, you’re dealing with binary Protobuf payloads and HTTP/2 multiplexing that most logs and traces simply weren't designed to handle. Speedscale flips the switch by decoding Protobuf directly into human-readable JSON in real-time. Get the speed of gRPC with the visibility of REST.

Ever tried searching your entire production stack for one user? Getting paged at 3 AM is bad enough. It’s worse when you only have a single username and zero visibility into what’s actually happening across your microservices. With Speedscale, you can perform full-text searches across every API call and database interaction in real-time. Stop guessing and start debugging with total context.

Learn more at speedscale.com.

Testing with production traffic doesn't have to be a security risk. Engineers often avoid production data because of sensitive info like passwords, tokens, and PII. But legacy test data management is too static for modern, fast-changing payloads. Enter the Speedscale Streaming DLP Engine. It automatically detects and redacts sensitive data in real time as it's captured from your environment. You get the realism of production traffic without the risk of a data breach.

Learn more: speedscale.com.

Learn more: speedscale.com.

Learn more at speedscale.com or read more at https://speedscale.com/blog/peeking-under-the-hood-with-claude-code/

While analyzing network traffic, we found huge amounts of telemetry including chat snippets, being sent to statsig.anthropic.com. The irony? Statsig was recently acquired by OpenAI. In this video, we use proxymock to intercept the traffic and show you exactly what’s being sent from your terminal to Anthropic (and technically, OpenAI’s infrastructure).

Everyone is talking about Claude Code, but few people understand the machinery running in the background. Today, we’re opening up the terminal to see how Anthropic’s coding agent manages state, runs tests, and fixes its own bugs. From the Model Context Protocol (MCP) to its unique React-based terminal UI, find out what makes Claude Code the most "senior" feeling AI assistant on the market.