Ask Claude Code to do a task... until it's done. --> Loop Engineering Recently, Boris Cherny (Head of Claude Code at Anthropic) shared: "I don't prompt Claude anymore. I have loops that are running. They're the ones prompting Claude and figuring out what to do. My job is to write loops." That perfectly describes where AI development is heading. We've gone through an interesting evolution: Prompt Engineering Crafting better prompts to get better responses. ⬇️ Context Engineering Giving AI the right knowledge, files, memory, and documentation so it can make better decisions. ⬇️ Harness Engineering Building the orchestration layer that decides which model to use, which tools to call, when to retry, and how different agents work together. ⬇️ Loop Engineering (where we're headed) Define the goal once. Let AI plan, execute, verify, fix, and repeat until the task is complete. A well-designed loop isn't just "AI running repeatedly." It's built on five key components: 1: Automation The loop runs independently. Once you define the objective, it keeps progressing without constant human intervention. 2: Parallel Agents Different agents work on different parts of the problem simultaneously—one writes code, another researches, another reviews or tests. 3: Skills Agents aren't limited to chatting. They can use tools, run code, call APIs, search documentation, execute commands, and interact with your development environment. 4: Context Every decision is grounded in the right information—your codebase, project docs, previous attempts, memory, logs, and requirements. 5: Connectors Loops become truly useful when connected to real systems like GitHub, Slack, databases, cloud platforms, CI/CD pipelines, and other services. Goal ↓ Plan ↓ Execute ↓ Verify ↓ Fix ↓ Repeat ↓ Done That's why tools like Claude Code feel fundamentally different. You're no longer spending your time writing prompt after prompt. You're designing systems that can keep working, even while you're away. Maybe the next must-have AI skill isn't prompt engineering. Maybe it's Loop Engineering. Who knows, a new 'Engineering' term is just around the corner!

This one is for all the overthinkers out there.

Meet Notto

A small creature made entirely out of thought.

Every mind leaves things behind: unfinished ideas, buried memories, questions we never answered. Notto is what happens to those traces.

Its personality lives entirely in its head, because that's where the thinking happens.

Calm, curious, overwhelmed, or inspired, it's always shifting, never twice the same.

The body stays constant underneath it all: quiet, grounded, unmoved. That contrast is the whole character. One part of Notto never changes. The other part never stops.

The process:

It started with just a silhouette. Every mascot you actually remember, you remember as a shape first. Before color, before detail, before anything else.

If the outline alone can't carry the identity, nothing else will save it. So that's where the exploration began: stripping the character down until the silhouette could stand on its own.

From there, the system built itself around one rule: the body stays constant, the head doesn't. The head is a living visualization of thought: every line is a memory, an idea, a question, a curiosity. Calm settles into smooth, rounded orbits. Curiosity pulls the lines outward. Overwhelm tightens them into sharp angles. Inspiration lets them glow and wander into color.

Every thought leaves a shape. Explore the full project here: https://www.recraft.ai/project/ebf1d1f9-f0f0-4c53-bf0d-90d6ce1b05a7

Same body. Same rules. Completely different headspace.

From left to right: calm and dreaming, scattered and overthinking, wired and overstimulated, quietly content, anxious and closed off, alert and searching. The body never moves. The head never lies.

This is the whole system in...