Reasoner-Executor-Synthesizer: Scalable Agentic Architecture with Static O(1) Context Window

TL;DR

The paper introduces RES, a scalable agentic architecture for LLMs that achieves constant token complexity regardless of dataset size, significantly reducing hallucinations and token costs in retrieval-augmented tasks.

Contribution

RES architecture separates reasoning, retrieval, and synthesis, using fixed-size summaries to ensure O(1) token complexity and eliminate data hallucination in large-scale retrieval tasks.

Findings

RES achieves O(1) token complexity with dataset size.

RES maintains low token cost (~1,574 tokens) across datasets from 42,000 to 16.3 million articles.

The architecture effectively eliminates data hallucination by design.

Abstract

Large Language Models (LLMs) deployed as autonomous agents commonly use Retrieval-Augmented Generation (RAG), feeding retrieved documents into the context window, which creates two problems: the risk of hallucination grows with context length, and token cost scales linearly with dataset size. We propose the Reasoner-Executor-Synthesizer (RES) architecture, a three-layer design that strictly separates intent parsing (Reasoner), deterministic data retrieval and aggregation (Executor), and narrative generation (Synthesizer). The Executor uses zero LLM tokens and passes only fixed-size statistical summaries to the Synthesizer. We formally prove that RES achieves O(1) token complexity with respect to dataset size, and validate this on ScholarSearch, a scholarly research assistant backed by the Crossref API (130M+ articles). Across 100 benchmark runs, RES achieves a mean token cost of 1,574 tokens regardless of whether the dataset contains 42,000 or 16.3 million articles. The architecture eliminates data hallucination by construction: the LLM never sees raw records. KEYWORDS LLM agents; agentic architecture; hallucination elimination; token optimization; context window; retrieval-augmented generation; deterministic execution; scholarly metadata; Crossref API; O(1) complexity.