GrantedAI.com is an AI-native grant writing assistant that utilizes advanced natural language processing (NLP) techniques and Large Language Models (LLM) to draft compelling and well-structured grant proposals. Here's a snapshot of how Granted operates:Our team, comprising professional grant writers, grant writing instructors, and Data Scientists, curates and compiles a comprehensive database of successful grant proposals, grant writing literature, and industry best practices. These data is then used to fine tune LLMs like GPT-3.5-turbo, Mistral 7b variants, and smaller BERT-based transformer models to complete different stages of the grant writing process. Clients provide some project information, such as objectives, high-level methodology, expected outcomes, and the funding agency's mission and goals. This information allows Granted to customize the generated text to meet specific needs and funding opportunity requirements.Granted's agents analyze the unique aspects of the project along with the funding agency's objectives. The AI then crafts text for the grant proposal that communicates the details of the proposed project while also ensuring alignment with the funding agency's own mission and priorities.

As a Partner at SciTech Strategies, I help to direct our research intelligence and science of science efforts. My work focuses on developing novel analytical frameworks and AI-driven tools to map the evolving landscape of scientific research, identify emerging areas of innovation, and predict the trajectory of research communities.Our work enables exploration of the research landscape at multiple levels, from broad domains down to specific concepts and discoveries, and through both commercial and academic lenses. This system is now being used by leading universities, funders, and government agencies to inform strategic decision making and optimize research portfolios.Another aspect of my role is fostering collaboration at the cutting edge of the science of science. I've initiated and led several high-impact partnerships, bringing together multidisciplinary teams from academia and industry to tackle ambitious research questions. For example, our ongoing collaboration with the Center for Security and Emerging Technology (CSET) and the University of Michigan has yielded groundbreaking insights into the drivers of scientific creativity and the mechanisms of translational research. This has allowed us to develop a novel quantitative framework for understanding and catalyzing the journey from basic research to real-world impact.We have been experimenting with generative AI and large language models like GPT-4 and Mistral 7b to understand how they can be used to automatically synthesize key findings across large bodies of research, generating accessible summaries tailored to different stakeholder groups on-demand. While still early, this work has shown great promise for democratizing access to scientific knowledge and evidence-based decision making.

As the Director of Research Intelligence & Analytics, my recent work focuses on developing a quantitative framework for the identification and characterization of translational research and the use of “grounded generation” in Large Language Models for text inference in domains where truthfulness and trustworthiness are paramount. For instance in Informed Consent documents. My work has led to the development of strategic models that identify scientific innovation and predict the growth of research communities as well as their interdependencies. I am proud to have helped produce the first quantitative definition of translational research (publication in progress), a significant advance in translational science and a valuable tool for program planning and evaluation. Recently, my team and I have been leveraging large language models such as GPT-4, Llama-2, Mistral, and Claude3 to approach ad hoc multi-document summarization in plain language. Our use of local models is important for communicating nuanced research outcomes to program managers, policy makers, and the public while remaining protective of confidential information.Another key area of my work involves the automatic identification and characterization of research expertise. This framework is currently being utilized by the University of Michigan Medical School for faculty recruitment and retention decision support and it has provided quantitative structure for the Medical School's most recent 5-year research strategic plan.