(20 Mar 2025) Research Solutions, Inc. announces significant enhancements to its AI product, Scite Assistant, that will help enterprise customers get more accurate responses to complex research questions and tasks.
Scite differentiates itself from other AI tools by pairing proprietary citation data and exclusive access to scholarly content with powerful AI models. The latest release deployed OpenAI o3-mini with reasoning, a substantially more powerful engine to drive Scite’s agentic AI product.
Unlike standard AI models that generate responses based primarily on pattern recognition, reasoning models are specifically trained to think problems through internally before answering, consider multiple potential approaches, evaluate the accuracy of different possible answers, and solve multi-step problems methodically.
“Reasoning models represent a significant advancement in AI technology for research applications,” said Josh Nicholson, Chief Strategy Officer at Research Solutions. “By combining our unique data and access to scholarly research, we’re delivering more accurate information synthesis, better identification of nuance and conflicting evidence, and higher-quality outputs than any other AI research tool on the market today.”
The new model also features customizable reasoning intensity, allowing users to control how deeply the model thinks problems through before generating responses. Along with existing settings—such as specifying journals, selecting publication date ranges, and scoping inputs to pre-selected article collections—this gives researchers a unique solution to accelerate their work.
To help customers understand the significant improvements these updates bring, Research Solutions has prepared detailed comparison documents showcasing the differences in quality and capabilities: Download Before & After: Reasoning Model Outputs (PDF)
These documents compare outputs from the previous and new versions side-by-side, demonstrating the substantial quality improvements across various research scenarios.
The press release in full is here.
