In the ever-evolving landscape of research funding, scientists in the United States are increasingly challenged by budget cuts and growing competition. Dr. Adam Kiefer, a researcher at the University of North Carolina at Chapel Hill (UNC), has firsthand experience with this. He often dedicates countless hours to reformatting similar grant proposals for different agencies, transforming an eight-page National Institutes of Health (NIH) submission into a concise one-pager for the Department of Defense, only to restructure it again for the Defense Advanced Research Projects Agency (DARPA). Meanwhile, he discovers grant opportunities that are new to him but are expiring in databases he was previously unaware of.
Kiefer’s situation is not unique. In fact, researchers spend around 44% of their time on administrative duties, which detracts from actual research work, with proposal preparation consuming the most time. The task of identifying new grant opportunities and crafting corresponding proposals entails manually sifting through various platforms like Grants.gov, the NIH, National Science Foundation (NSF), and many additional resources. Early-career researchers face even greater challenges due to a lack of established networks or email lists from relevant agencies.
In response to these challenges, Amazon Web Services (AWS) collaborated with Kiefer to develop the Grant Research Opportunity Wizard (GROW). This open-source generative AI tool, available on the AWS Samples GitHub repository, is designed to expedite the proposal development process and is currently evolving to feature autonomous capabilities that will automatically identify relevant grant opportunities globally.
How the idea for GROW emerged
The inception of GROW stemmed from years of partnerships between the AWS Academic Research team and Kiefer, focusing on enhancing cloud-native research capabilities. Throughout these discussions, Kiefer frequently expressed his frustrations: the unending cycle of adjusting grant proposals and the struggle to uncover opportunities beyond well-known funders. Occasionally, these frustrations would result in abbreviated meetings or even cancellations due to imminent deadlines.
“The burden was overwhelming,” Kiefer reflected. “Daily, I was asked to integrate something new into my proposal—another section, another format. It dawned on me that there must be a more efficient approach.”
The AWS team recognized that these challenges extended beyond Kiefer’s individual experience and indicated a technological gap that needed addressing.
This insight sparked an innovative idea. Kiefer collaborated with the Academic Research business development and solutions architect teams to create the initial version of GROW, concentrating initially on proposal development. “The ideas they returned with exceeded my expectations. It was exhilarating,” Kiefer remarked.
Streamlining NIH proposals
The outcome was GROW version one, which was released as open-source on the AWS Samples GitHub repository in late 2024. This version helped researchers draft NIH proposals more efficiently by utilizing generative AI. It processed a researcher’s publications and draft materials to create proposal content aligned with NIH guidelines.
“As a researcher, formulating a project concept and drafting it on paper takes me at least three weeks,” explained Jack Fenwick, AWS business development leader. “Having a preliminary draft that aligns with my expertise and solicitation requirements—rather than starting from scratch—can save me weeks in developing a comprehensive proposal.”
Various research organizations nationwide, including prominent universities, deployed GROW version one as a pilot. Feedback from these institutions highlighted a significant gap: researchers required assistance not only in composing proposals but also in identifying new grant opportunities beyond traditional sources.
From proposal drafting to autonomous discovery
This input instigated a pivotal transformation. GROW version two transitioned from a generative AI that responds to user prompts to an agentic AI that autonomously identifies global funding opportunities tailored to a researcher’s expertise. This version leverages tools such as Amazon Bedrock, Amazon Bedrock AgentCore, and Amazon OpenSearch Service. It also includes safeguards to protect sensitive research data while ensuring that organizations maintain complete control over their research profiles and proposal content.
Upon integration into a research environment, GROW version two utilizes AI agents to autonomously scour funding databases multiple times a day, including resources like Grants.gov, various federal agency websites, and international opportunities such as Horizon Europe. Researchers can create profiles through CV uploads or by linking existing online profiles, allowing them to establish multiple profiles for differing research areas. They can personalize their searches by outlining their expertise, career stage, target topics, and preferred funding agencies.
Kiefer believes that this flexibility is crucial for his research endeavors. “I will ensure my profile highlights what is pertinent for this specific research line and then let the agentic AI do the work for me,” he stated.
By utilizing extensive knowledge bases, GROW matches opportunities to researcher profiles, publications, and expertise, uncovering grants that conventional searches might overlook and providing them with a matching score that indicates how well the opportunities align with their profiles. Once a researcher identifies a suitable opportunity, GROW streamlines the process by extracting best practices and guidelines for each funder, producing proposal templates tailored to specific requirements. This results in a significant head start in both seeking and crafting proposals.
Finding the right grant opportunities faster
For Kiefer, GROW has opened doors to opportunities previously inaccessible to him. “I have collaborators in Europe, yet I have limited knowledge of the European funding landscape,” he admitted. “GROW can navigate that system and identify potential opportunities for me—I wouldn’t even know where to start if I had to do it manually.”
Moreover, GROW unveils funding avenues in areas that researchers might not typically consider. Kiefer understood that his eye-tracking studies on basketball performance could be relevant to military contexts concerning soldiers’ battlefield awareness. However, he felt unacquainted with military funding structures. “GROW bridges that gap,” he claimed. “Without GROW, I would spend countless hours trying to uncover these options on my own, and I might still miss some.”
Yet, increased discovery does not mean chasing every possible opportunity. GROW enables researchers to concentrate their efforts strategically. “What GROW provides is precision,” Kiefer clarified. “It casts a wide net but retrieves high-probability matches, allowing you to channel your attention toward the prospects most likely to yield success.”
This strategic focus translates to more effective efforts rather than increased workload. “I am not interested in writing more grants,” Kiefer added. “My goal is to devote more time to crafting proposals where my success likelihood is considerably higher, rather than wasting time pursuing opportunities that I may have stumbled upon.”
For Fenwick, this shift signifies a broader perspective: “It significantly widens the lens from a narrow viewpoint, shaped by personal experience, to a much broader recognition of what is available and achievable.” Preliminary estimates suggest that GROW could enhance researchers’ efficiency by up to 30%, allowing them to redirect time toward refining their grant-seeking strategies and creating more competitive submissions.
Open source for any research organization
This efficiency boost is set to benefit the wider research community soon. GROW version two is anticipated to be released as open-source on the AWS Samples GitHub repository in early 2026. By making GROW available as an open-source prototype, AWS aims to empower research organizations of all scales to access the same discovery and proposal tools, facilitating researchers in focusing on their work and enhancing research output across various institutions. Universities, government laboratories, nonprofits, and research institutes will have the capability to deploy, customize, and extend their functionalities.
These advancements mark just the beginning of GROW’s potential. Its roadmap now encompasses not just grant discovery but also proposal submission, budget development, and post-award management.
Ultimately, this collaboration exemplifies the transformative outcomes achievable when researchers and technology developers unite to confront shared obstacles. “This is merely one instance of what can occur when an academic researcher collaborates with an industry giant like AWS,” Kiefer noted. “It can fundamentally alter how we perceive and address the manual, time-heavy inefficiencies prevalent in the research ecosystem, paving the way for improvements for the broader community.”
Contact AWS today to discover more about how we support higher education and research.
Read related stories on the AWS Public Sector Blog
