Prompt vs Loop vs Graph Engineering: Three Levels of Control in AI Development

On July 29, 2026, MarkTechPost broke down three fashionable terms in AI engineering — prompt engineering, loop engineering and graph engineering — as three levels of control nested inside one another, not as competing approaches.

Three levels of control

Prompt engineering, loop engineering and graph engineering are not rival techniques but different units of control stacked on top of each other. A prompt governs one model response, a loop governs the behavior of a single agent, and a graph governs the organization of multiple agents. Each higher layer keeps the one beneath it: the prompt does not disappear when a loop is built around it — it simply stops being something people type by hand.

The terms entered practice at different times. Prompt engineering is well established. Loop engineering entered the vocabulary in late 2025 and dominated developer discussions until June 2026. Graph engineering followed about six weeks later and is so far the least settled of the three.

What the stack is made of

The full progression has five steps: prompt → context → harness → loop → graph. Practitioners named each of them before it made it into vendor documentation.

• Prompt engineering — writing the instruction for a single call; Anthropic advises splitting the system prompt into sections (background, instructions, tool use, output description) using XML tags or Markdown headings
• Context engineering — according to Anthropic's description, the natural continuation: the question shifts from "which words to choose" to "what configuration of tokens should be in the window at all"
• Harness engineering — the environment of a single agent: files, tools, memory, feedback
• Loop engineering — one floor up: how the system repeatedly observes, acts, verifies and recovers
• Graph engineering — graph-based orchestration of many agents

A June 2026 arXiv paper on agentic AI in construction engineering (the Buildrix project) explicitly lays out the same chain: prompt, then context, then harness, then loop.

Where a single prompt breaks down

Prompt engineering assumes a human is present at every iteration: the prompt is written, the model answers, the output is evaluated, the prompt is rewritten. That assumption is exactly what collapses. High volume, multi-step tasks, no human available to evaluate, results that feed automatically into the next step — any one of these conditions is enough for a single prompt to stop being sufficient.

The prompt itself does not get worse; the surrounding conditions change. And prompt engineering does not vanish inside the higher layers: according to Anthropic's report on multi-agent research, prompts were the main lever for fixing coordination failures. Early versions spawned 50 subagents for simple queries, and the fix came from prompting, not from topology.

"Prompt engineering was the primary lever for improving coordination failures," says

Anthropic's research report on multi-agent systems.

What loop engineering is

Loop engineering treats a coding agent as a brute-force tool for searching for solutions, and treats the craft as designing the goal, the tools and the loop itself, not just the prompt. The term went mainstream in June 2026 after a widely shared post whose author urged engineers to stop "prompting" coding agents and start designing loops that prompt them instead. The same week, the Claude Code team at Anthropic described the same shift from the stage.

The most detailed public breakdown identifies five primitives and a sixth element that ties them together: automations (a schedule or event that triggers discovery and triage without supervision), worktrees (isolation so that parallel agents do not edit the same files) and others.

What this means

The three terms describe not a fad but three different units of control — a model response, an agent loop and an organization of agents. Knowing which layer a task sits on determines what to design: the words of a prompt, a feedback loop or a coordination graph.

Frequently asked questions

How does loop engineering differ from prompt engineering?

A prompt governs one model response, while a loop governs an agent's behavior cycle: how it observes, acts, verifies and recovers. A loop does not replace the prompt; it is built on top of it.

When did the term loop engineering appear?

According to MarkTechPost, loop engineering entered the lexicon in late 2025 and dominated developer discussions until June 2026, when it went mainstream after a widely shared post.

What is graph engineering?

Graph engineering is the newest of the three terms, appearing about six weeks after loop engineering. It means graph-based orchestration of many agents; the term's origin is not yet settled and overlaps with the older meaning from knowledge graphs.