Arcee AI's nac coordinates long engineering tasks by separating planning from execution. A central AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → assigns work to workers but cannot itself edit files or run commands. Each worker finishes with a handoff describing its results; nac retains that handoff and discards the worker's execution context windowThe maximum amount of text a model can consider at once — its working memory for the current conversation or task.Full definition →. When the work resumes, a fresh worker receives the retained history for that workstream and selected handoffs from others. Arcee says it built nac this way so progress can accumulate without forcing every intermediate tool response and failed attempt through one growing conversation. Architecture account.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
agent harness — The scaffolding around a model that turns it into a working agent — the loop, the tools it can call, and the rules for when to stop.
