Loop engineering represents a fundamental shift in how AI systems work: instead of humans writing new instructions each time an AI model responds, engineers now design automated systems that repeatedly observe, decide, act, and verify their own progress toward a goal. This transition, formalized in June 2026, marks the moment when manual prompting became impractical for real-world AI workflows.

What's the Difference Between Prompts and Loops?

Prompt engineering optimizes a single exchange. You write an instruction, the model responds, you read the output, and if it's wrong, you write a better instruction. It is, structurally, a conversation where you hold the pen the entire time.

Loop engineering optimizes something bigger: the system that runs that conversation on your behalf, repeatedly, without you reading every reply. The model stops being the person you're talking to and becomes a component inside something you designed. A prompt engineer is judged on one good answer. A loop engineer is judged on whether a system can be trusted to run unattended for an hour, a day, or a sprint.

The contrast with the workflows most teams already have in production is sharp. A fixed pipeline runs step A, then step B, then step C, in that order, every time. A loop is dynamic. The agent might go from A to B, discover B didn't work, revise its approach, and only then move to C, or it might loop back to A entirely.

Who's Driving This Change?

Three industry leaders made the same argument within weeks of each other in June 2026, signaling that loop engineering had arrived as a discipline. Boris Cherny, who built Claude Code at Anthropic, said in an interview: "I don't prompt Claude anymore. I have loops that are running. They're the ones that are prompting Claude and figuring out what to do. My job is to write loops." Nvidia's Jensen Huang told developers something structurally identical: "Nobody writes prompts anymore. The new job is to write and handle loops." Peter Steinberger, the founder of PSPDFKit and creator of OpenClaw, made the same argument to millions of viewers in a single day.

"I don't prompt Claude anymore. I have loops that are running. They're the ones that are prompting Claude and figuring out what to do. My job is to write loops," said Boris Cherny.

Boris Cherny, Creator of Claude Code at Anthropic

The name for the new discipline itself has a documented origin. It was popularized in June 2026 by Addy Osmani, engineering lead at Google Chrome, synthesizing ideas from Boris Cherny at Anthropic and Peter Steinberger, founder of PSPDFKit/Nutrient.

How Did We Get Here From Prompt Engineering?

The progression from prompts to loops didn't happen overnight. Research on agentic software engineering traces a direct lineage through four distinct phases, each solving real failures the previous layer couldn't address:

• Prompt Engineering: Focuses on how a task is expressed to the model, optimizing a single instruction and response.
• Context Engineering: Determines which instructions, data, tools, and intermediate information should be made available to the model.
• Harness Engineering: Connects the model to an executable environment containing files, software, memory, and feedback mechanisms.
• Loop Engineering: Extends these ideas by defining how the system repeatedly observes, acts, verifies, and recovers until achieving a goal.

Prompt engineering ran out of room once models needed more information than a clever sentence could carry. Context engineering ran out of room once agents needed to actually do things, not just know things. Harness engineering ran out of room the moment a task took longer than one action to finish, which is most real work.

Most modern agent loops trace back to the ReAct pattern (Reason and Act), introduced in research from Princeton and Google. The idea: interleave reasoning steps with action steps. The model thinks out loud, takes an action, sees what happened, thinks again, and acts again. What changed in 2026 wasn't the pattern. It was the moment enough senior engineers admitted, on the record, that manual prompting couldn't keep pace with what the pattern demanded.

Is Prompt Engineering Dead?

Not exactly. Prompt engineering got demoted, but it didn't disappear. That framing is more accurate than the "prompt engineering is dead" headlines circulating this year, and it matters for how organizations plan hiring and training. According to Gartner, "context engineering is in, and prompt engineering is out".

Here's the version that actually holds up under scrutiny: prompt engineering didn't disappear, it moved inside the loop. A loop still needs prompts, it just doesn't need a human writing a new one every time the model responds. If your team spent 2024 and 2025 building prompt libraries, that work isn't wasted. It's raw material. The prompts you refined become the reusable components a loop calls on when it decides an action is needed, not the thing a human retypes every morning.

What Impact Does Loop Engineering Have on AI Performance?

The effect on output quality, when it's engineered well, is not subtle. A loop is a repeating cycle with four moving parts: it observes something, decides what to do, acts on that decision, and checks whether the result moved it closer to a goal. That difference is not cosmetic. It's the difference between an agent that fails the moment reality deviates from the script, and one that notices the deviation and adjusts.

Enterprise workflows deviate from the script constantly. A document is malformed, an API times out, a customer's request doesn't match the expected pattern. A pipeline breaks on that. A loop absorbs it. Andrew Ng highlighted a case where ChatGPT 3.5 was used in an agentic workflow instead of a simple one-shot response, and its coding accuracy jumped from 48.1% to 95.1%. Same model. Same underlying capability. The only thing that changed was the architecture.

How to Transition Your Team From Prompt Engineering to Loop Engineering

• Audit Existing Prompts: Inventory the prompt libraries your team built in 2024 and 2025. These become reusable components inside loops, not standalone instructions.
• Design Feedback Mechanisms: Build systems that let your AI agents observe the results of their actions and adjust course. This is the core of loop engineering.
• Define Clear Termination Conditions: Specify when a loop should stop, whether because a task is complete, a stopping criterion triggers, or the agent determines it can't proceed further.
• Test Unattended Execution: Shift your success metrics from "Did this one response work?" to "Can this system run for an hour without human intervention and still achieve its goal?"

The shift from prompt engineering to loop engineering is not a rebrand of an old idea with a trendier name. It's a shift in what the engineer is responsible for, from the quality of one response to the reliability of an entire execution process. For enterprise AI teams trying to move agents from pilot to production, understanding this distinction determines whether your 2026 roadmap is aimed at the right target.