Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

ResearchPilot: A Local-First Multi-Agent System for Literature Synthesis and Related Work Drafting

Peng Zhang

Abstract

ResearchPilot is an open-source, self-hostable multi-agent system for literature-review assistance. Given a natural-language research question, it retrieves papers from Semantic Scholar and arXiv, extracts structured findings from paper abstracts, synthesizes cross-paper patterns, and drafts a citation-aware related-work section. The system combines FastAPI, Next.js, DSPy, SQLite, and Qdrant in a local-first architecture that supports bring-your-own-key model access and remote-or-local embeddings. This paper describes the system design, typed agent interfaces, persistence and history-search mechanisms, and the engineering tradeoffs involved in building a transparent research assistant. Rather than claiming algorithmic novelty, we present ResearchPilot as a systems contribution and evaluate it through automated tests and end-to-end local runs. We discuss limitations including external API rate limits, abstract-only extraction, incomplete corpus coverage, and the lack of citation verification.

ResearchPilot: A Local-First Multi-Agent System for Literature Synthesis and Related Work Drafting

Abstract

ResearchPilot is an open-source, self-hostable multi-agent system for literature-review assistance. Given a natural-language research question, it retrieves papers from Semantic Scholar and arXiv, extracts structured findings from paper abstracts, synthesizes cross-paper patterns, and drafts a citation-aware related-work section. The system combines FastAPI, Next.js, DSPy, SQLite, and Qdrant in a local-first architecture that supports bring-your-own-key model access and remote-or-local embeddings. This paper describes the system design, typed agent interfaces, persistence and history-search mechanisms, and the engineering tradeoffs involved in building a transparent research assistant. Rather than claiming algorithmic novelty, we present ResearchPilot as a systems contribution and evaluate it through automated tests and end-to-end local runs. We discuss limitations including external API rate limits, abstract-only extraction, incomplete corpus coverage, and the lack of citation verification.
Paper Structure (26 sections, 1 figure, 3 tables)

This paper contains 26 sections, 1 figure, 3 tables.

Table of Contents

  1. Introduction
  2. Problem Setting
  3. Related Work
  4. Retrieval-Augmented Generation
  5. Multi-Agent LLM Systems
  6. LLM Orchestration Frameworks
  7. Scholarly Search and Review Tools
  8. System Overview
  9. Agent Pipeline
  10. SearchAgent
  11. ExtractionAgent
  12. SynthesisAgent
  13. WriterAgent
  14. Stage Contracts
  15. Orchestration and Interface Design
  16. ...and 11 more sections

Figures (1)

  • Figure 1: Architecture of ResearchPilot. The frontend communicates with the FastAPI backend, which orchestrates a four-stage agent pipeline and persists reports in SQLite and Qdrant.