UBIT is expanding BullsAI access and use cases across campus, including additional faculty onboarding, new pilot departments, and deeper integration with institutional workflows. Access is still broadening, with 30 UB faculty and limited students currently using the tool.

UBIT is in the process of rolling out more features such as multi-step agentic reasoning and the ability to deploy custom assistants.

UBIT deployed an immersive 3D wall that is being piloted by media studies faculty, with additional faculty onboarding planned. The technology supports experiential learning by making complex, spatial, and data-rich concepts easier to explore collaboratively.

• Instructional value: Allows students and researchers to examine scale, depth, motion, and relationships that are difficult to understand through flat screens or static models.
• Collaboration: Supports active inquiry, simulation, design review, and problem-solving in a shared teaching environment.
• Research and teaching alignment: Connects advanced research infrastructure with high-impact learning experiences. 

UBIT launched GitHub Copilot developer tool pilots across select technical teams, overseen by UBIT's AI lead. The pilot is continuing through Spring 2027 to help the university build practical experience with AI-assisted software development, evaluate governance and responsible-use considerations, and better understand the costs, benefits, and long-term role of AI coding tools in higher education technology environments.

Initial survey feedback has been generally positive, with participants citing potential benefits such as reducing repetitive work, accelerating prototyping, improving troubleshooting and debugging, and allowing developers to focus more time on design, architecture, and problem-solving activities. Early usage data also shows strong engagement with advanced AI capabilities, with approximately 69% of active customers exploring agent-based features beyond basic code completion. 

• Pilot community: Approximately 50 developers within UBIT and 15 others around the university.
• Model access: Access to the latest ChatGPT and Claude models for coding and related work. 

UBIT launched BullsAI, UB’s enterprise AI platform, providing secure, institutionally governed access to large language models. The implementation involved a cross-functional team and pilot groups that include CATT, English faculty, UB’s School of Management, and UBIT's Innovation and Enterprise Application Services teams. The Provost’s AI & Society funding supported the effort, with UBIT providing technical, functional, and project management resources.

• Platform hardware: Four NVIDIA DGX B200 servers, 32 NVIDIA Blackwell B200 GPUs with 5.7 TB GPU memory, 1,400 total system CPU cores, 12 TB total system memory, and a 500 TB flash storage array. 
• Pilot community: Approximately 150 chat customers and 50 developers.
• Model access: BullsAI currently runs 15 different open-weight LLMs accessible via chat and API key. 
• Token processing: BullsAI processes 5 billion tokens per month securely on its own hardware.

UBIT piloted Microsoft 365 Copilot Premium licenses with over 600 campus stakeholders across Exchange, Word, Excel, PowerPoint, Teams, and related Microsoft 365 content. The pilot explored practical use cases such as summarizing meetings and chats, finding historical project information, drafting and revising guidance, consolidating spreadsheet data, and supporting technical work such as query writing and configuration file generation.

In a recent survey, 81% of UB staff and faculty (8% of total) respondents noted improvement in productivity when using MS Copilot Premium at UB. 63% said Copilot Premium saves them time, 61% noted the tool synthesizes a large amount of information, 58% said it improves their writing quality, and 50% said it helps them get started on tasks. 

Between June-July 2026, UB’s Microsoft Copilot tools (Copilot Premium, Copilot Chat and other agents) conducted over 146,000 activities.