Scinaut: one AI research assistant with data & novelty analysis

The idea to build this AI research website came from my own frustrating experience working on research projects and reviewing literature. Locating papers, reading through studies, and organizing references require jumping between dozens of separate tools. Simply extracting key arguments and formatting citations eats up hours of work. What’s more, generic AI tools on the market lack an understanding of rigorous academic writing logic and often generate unreliable, imprecise content. That’s why we set out to build a dedicated tool for researchers. Beyond assisting with literature review, it accurately identifies the novel contributions of papers, objectively exposes flaws within experimental data, and delivers tiered experimental innovation ideas powered by professional Design of Experiments (DOE) tools. We aim to address the core pain point of low efficiency across the entire research workflow: academic resources are scattered and hard to source; manually sorting core arguments and experimental data from lengthy papers is extremely time-consuming; formatting each reference entry one by one after finishing a paper creates disjointed, cumbersome workflows. This platform unifies all essential capabilities in one place, including literature retrieval, in-depth paper analysis, novelty & data evaluation, and interactive Q&A for paper-based learning — eliminating the need to switch constantly between multiple tools. We revised our product vision repeatedly throughout development. Initially, we only planned a basic paper summarization feature. After gathering genuine feedback from students and professional researchers during internal testing, we iterated and expanded the product into a full end-to-end research workflow. We upgraded our literature parsing model to clearly isolate experimental datasets and conflicting scholarly viewpoints, while adding support for all major domestic and international citation standards. Finally, we streamlined the interface to lower the learning curve, enabling complete beginners to master the tool quickly.

Maker

📌

2mo ago

The "no credit card" free tier is a smart move for a tool aimed at researchers, who already deal with enough friction jumping through institutional paywalls. Curious how the Q&A handles multi-step reasoning across dense experimental sections.

Uploaded a dense perovskite paper and the summary actually pulled out the key synthesis details I needed for a meeting prep. Genuinely useful for anyone who reads journals all day.