SciFi: A Safe, Lightweight, User-Friendly, and Fully Autonomous Agentic AI Workflow for Scientific Applications

TL;DR

This paper introduces SciFi, a safe, lightweight, and user-friendly autonomous AI framework designed for scientific research, enabling reliable, end-to-end automation of structured tasks with minimal human oversight.

Contribution

The paper presents a novel autonomous agentic framework combining safety features, a three-layer agent loop, and self-assessment to improve deployment reliability in scientific workflows.

Findings

Supports end-to-end automation of scientific tasks

Ensures safe operation with self-assessment and stopping criteria

Reduces human intervention in routine scientific workflows

Abstract

Recent advances in agentic AI have enabled increasingly autonomous workflows, but existing systems still face substantial challenges in achieving reliable deployment in real-world scientific research. In this work, we present a safe, lightweight, and user-friendly agentic framework for the autonomous execution of well-defined scientific tasks. The framework combines an isolated execution environment, a three-layer agent loop, and a self-assessing do-until mechanism to ensure safe and reliable operation while effectively leveraging large language models of varying capability levels. By focusing on structured tasks with clearly defined context and stopping criteria, the framework supports end-to-end automation with minimal human intervention, enabling researchers to offload routine workloads and devote more effort to creative activities and open-ended scientific inquiry.