AI Launch

AI Launch is a competitive program for UVA research teams. Selected teams get expert mentorship, a hands-on workshop, and MAP Lab engineering support to prototype agentic AI systems that transform health and life sciences research. Letters of Intent are due September 29, 2026.

Overview 

AI Launch invites research groups across UVA to think about redesigning how health and life sciences research is done by developing agentic AI systems that augment and automate key parts of the scientific research process. Through a combination of expert mentorship, hands-on workshops, and technical consultation – as well as development and infrastructure support from UVA's Mobile, AI, and People (MAP) Lab and other collaborators – participating teams will co-design and prototype agentic AI systems tailored to improving their research programs.  

The envisioned systems can assist with hypothesis generation, experimental design, computational workflows, literature synthesis, laboratory workflows, clinical research, and scientific communication and teamwork. Rather than funding individual research projects, AI Launch provides expertise, engineering support, and computational resources needed to build a working prototype that positions the teams for larger external funding opportunities. 

What You Might Build 

The best proposals begin with an existing research challenge or workflow that can be transformed through AI-augmented approaches. The prototypes envisioned in AI Launch are agentic AI systems: software that can reason, plan, use external tools, and coordinate complex scientific workflows across multiple steps with appropriate human oversight. Depending on the application, these systems may consist of a single AI agent or multiple collaborating agents. Some examples: 

• Run a virtual team of AI agents in the roles of a real lab to pressure-test an experimental design before anyone touches the bench. 
• Read your own results and the literature to propose and rank the next experiment. 
• Run a standard analysis pipeline from a plain-language request, so a bench scientist is not waiting on a computational collaborator. 
• Turn a written protocol into runnable steps for a liquid-handling robot, with deviation flags. 
• Pull structure out of clinical or observational data that now takes a person weeks to assemble. 

Who Should Apply 

Research teams from across UVA working in health, medicine, biology, neuroscience, public health, engineering, behavioral science, and related disciplines are encouraged to apply. No prior experience developing AI agents is required. What matters is that your lab has a well-defined research challenge or workflow that could benefit from AI agents, along with an established research program capable of evaluating and adopting the resulting prototype. 

A single PI may apply, although collaborative teams are encouraged. Teams may include postdoctoral fellows, staff, students, and external collaborators. Note: Workshop participation may be limited to a subset of team members depending on overall interest and available space. 

How It Works 

The program runs in three stages around a workshop: 

Stage 1

Letter of Intent. Name the problem and how the lab handles it now, the area it falls in, where an agent could help, and the support you need most. (LOI submission form below.)

Cohort selection. An AI Launch review committee with expertise in artificial intelligence, health and life sciences, and AI deployment will review the Letters of Intent and invite a curated cohort of teams to participate in the workshop. In addition to proposal quality, the committee will consider the opportunity to assemble a cohort representing diverse application areas, research domains, and technical needs to maximize collaboration and shared learning. Because workshop capacity is limited, not all applicants will be invited. 

Stage 2

Workshop (2 days). Participants will hear from experts developing AI agents for scientific discovery, laboratory automation, digital health, and clinical research. Interactive design sessions will help each team define the scope, architecture, evaluation strategy, and technical roadmap for its proposed system. 

Proposal development. During the proposal development phase, teams are encouraged to work with the MAP Lab to strengthen their proposals. The MAP Lab can help refine project plans and connect investigators with complementary expertise across UVA (e.g., the Link Lab for projects involving robotics and physical AI). Teams are encouraged to establish these collaborations prior to award decisions whenever possible.  

Full proposal (optional). After the workshop, teams may submit a full proposal for the competitive funding phase. Proposals should describe the current research workflow (baseline), the proposed agentic AI system, required data and computational resources, the technical development plan, an evaluation strategy with a comparator and primary metric, plans for human oversight, project team and roles, safety and regulatory considerations (including IRB and biosecurity, where applicable), and a strategy for future external funding, including potential grant mechanisms (e.g., NIH, NSF, ARPA-H, foundations, or industry partnerships) that the prototype is intended to support. 

Stage 3

Prototype Development. Two to four teams will be selected to work with a dedicated MAP Lab development team over a three- to six-month period to build and evaluate their agentic AI prototype. Support may include AI engineering, software development, cloud infrastructure, compute and model credits, and access to specialized expertise (e.g., robotics and physical AI). The program concludes with a demonstration day where teams showcase their prototypes and discuss next steps toward external funding. All participating labs retain their project designs and intellectual contributions. 

Before development begins, projects involving identifiable human data or raising dual-use, export control, or biosecurity considerations must obtain all required approvals, including IRB and other applicable institutional reviews. 

Selection Criteria 

• Impact: The measurable impact for the proposed AI system to meaningfully improve the research process compared to current practice. 
• Fundability: The potential for the resulting prototype to support competitive external funding and establish a sustainable research program. 
• Feasibility: The technical and organizational readiness to successfully develop, evaluate, and deploy the prototype using available expertise and infrastructure at UVA. 

Every proposal must also address human oversight, failure detection across long workflows, safety, and data governance. 

Support and Funding 

• The program provides technical expertise, engineering support, and development resources. No direct funds are awarded to participating laboratories. 
• All teams selected for the cohort will participate in the AI Launch workshop and receive access to domain-matched experts, proposal development support, and technical consultation from the MAP Lab and associated resources. 
• The two to four selected projects will receive a dedicated MAP Lab development team — including Graduate AI Fellows, undergraduate software developers, and shared project management — along with access to computing resources, model credits, software infrastructure, and, where appropriate, specialized hardware. The level of support will be tailored to the scope and technical needs of each project. 
• AI Launch is supported by the Jefferson Trust and administered through AI Research @UVA. 

 Key Dates 

• Letters of Intent due: Tuesday, September 29, 2026
• Cohort invitations: Thursday, October 15, 2026 (after letter review)
• AI Launch workshop: November/December 2026 (2 days) 
• Full proposal due: Friday, January 15, 2027  
• Award notification: Monday, February 1, 2027 

Questions? 

Contact Jason Nabi