From Pilot to Platform: BU’s SAIL Wins Sloan Grant to Scale AI Access
Open-source framework to make generative AI more reproducible, secure, and accessible across higher education
Boston University’s Software & Application Innovation Lab (SAIL) at the Hariri Institute has been awarded a $400,000 grant from the Alfred P. Sloan Foundation to address a growing challenge in higher education: how to move from experimentation with artificial intelligence to systems that are reliable, secure, and broadly usable.
The project, Providing Reproducible and Equitable AI Access in Academia (PREAA), develops open-source tools and practices that help researchers deploy generative AI in line with academic standards of transparency, reproducibility, and rigor.
Recent advances in AI have rapidly expanded what’s possible across research, teaching, and administration. Yet, universities often face a disconnect between interest in these tools and the ability to implement them effectively.
“At BU—and across universities more broadly—there’s no shortage of innovative ideas from students, faculty, and staff,” says Asad Malik, Principal AI Architect, PREAA. “But turning those ideas into stable, secure, and reliable systems remains a significant challenge. PREAA is about building the foundation that makes that possible.”
Building the Foundation
PREAA began as a pilot project supported by a $250,000 Sloan Foundation grant in 2024. Its aim was to develop a framework for AI applications tailored to academic settings using retrieval-augmented generation (RAG) – a method that improves large language models (LLMs) by grounding responses in retrieved information. Rather than building tools from scratch, the project set out to create a shared, open-source infrastructure to support diverse use cases while meeting institutional privacy and security requirements.
“We began with a focus on RAG tools but quickly discovered the ecosystem was much broader,” says Malik. “That led us to expand our scope to include more general-purpose open-source AI tools.”
A Year of Impact
Over the course of a year, the team developed an ecosystem that allows researchers to build, evaluate, and monitor AI-driven workflows across the full application lifecycle, from prototyping to deployment.
The pilot demonstrated strong demand across campus. Eighteen projects, spanning research, teaching, and administrative applications, were developed using the framework. Several projects were featured at BU Spark! 2025 Demo Day, illustrating how students could use the system to build their own AI-driven tools.
In research settings, the framework’s adaptability was showcased in multi-lab pilots where it provided a flexible open-source foundation for diverse applications, including data analysis and workflow automation.
PREAA proved accessible beyond technical disciplines. Faculty and students in fields, including Business, Chemistry, Brain Science, Biomedical Engineering, Electrical and Computer Engineering, and Law, were able to use the system without requiring deep expertise in machine learning or computer systems.
“The pilot showed that with the right infrastructure, researchers across disciplines could engage meaningfully with AI,” says Malik. “By shifting the focus from building one-off tools to applying them thoughtfully within a research context, PREAA lays the groundwork for subject matter experts to actively shape development.”
Streamlining Grant Discovery with AI
PREAA is also being adapted for administrative use to make it easier for faculty to find funding opportunities. The new AI-powered tool, called the Grant Navigation & Opportunity Scoring for Investigators & Scholars or Gnosis (Greek for knowledge), helps faculty identify grants by analyzing their CVs, Google Scholar profiles, or research summaries. Aggregating opportunities from federal, state, industry, and foundation sources, Gnosis gives researchers a curated starting point that matches their expertise. Gnosis is currently in beta and piloted with a focus group of faculty with a broader launch anticipated to the BU-wide community later this year.
“PREAA transforms how researchers discover funding,” says Yannis Paschalidis, Director, Hariri Institute for Computing. “Researchers can now rely on AI to search and prioritize funding opportunities that match their expertise, recent work, and, possibly, research plans described in an abstract, saving them considerable time spent on searching for suitable opportunities.”
From Experimentation to Infrastructure
The new Sloan grant will support PREAA’s transition from a pilot initiative to a more mature, sustainable, and community-driven platform. This next phase focuses on strengthening the system’s technical foundations while also addressing the organizational challenges of integrating AI into academic work.
“AI tools are quickly becoming central to the academic research enterprise, yet the barriers to building generative AI systems for individual projects are too high for many research groups,” says Joshua M. Greenberg, Technology Program Director at the Alfred P. Sloan Foundation. “PREAA is already helping to solve this problem locally at Boston University, and this next phase of work will help to transition it to a community-owned, accessible resource that benefits universities broadly.”
On the technical side, the project will enhance the platform’s core infrastructure, improving deployment pathways and incorporating capabilities for evaluation, tracing, and governance. These additions are intended to support more consistent and reproducible AI workflows, particularly as projects scale in complexity.
At the same time, PREAA will formalize collaboration practices between AI developers and domain experts. Because generative AI systems are inherently non-deterministic, effective use depends on coordination between those building systems and those applying them in specific fields. The project will develop shared workflows and guidelines to support this kind of interdisciplinary work.
“A big part about the next phase is having research groups take advantage of the findings from PREAA, be able to develop their own systems and contribute back to the community,” says Malik.
Building a National Community
While developed at Boston University, PREAA’s next phase aims to build a nationally connected academic ecosystem by providing open, adaptable tools and practices and expanding engagement with research software engineering groups across institutions.
“PREAA represents a step toward bridging the gap between AI experimentation and enduring academic impact,” says Malik. “The goal is to create systems that are not only innovative, but also reliable and sustainable over time.”
As universities continue to explore the role of AI in research and education, initiatives like PREAA point toward a model in which technical infrastructure, shared practices, and community engagement evolve together — making it possible to move from isolated experimentation to broadly supported, reproducible systems. Ensuring that researchers and institutions can freely adopt, adapt, and contribute to the ecosystem, all major PREAA resources, including codebases, templates, and documentation, are publicly available on GitHub under the MIT license: (https://github.com/hicsail/PREAA).