Let’s get personal

Imagine having a digital assistant that knows your corner of Yale inside and out. That’s now possible with Clarity Custom Agents, a new tool offered by Information Technology Services (ITS) that enables you to create personalized AI assistants trained on policies, documents, workflows, and much more. To learn about this new service and how to get started, Your Yale sat down with Michael Vaughn, AI Architect with the AI Platform Services team.

Why is Yale offering custom agents in Clarity?

We kept hearing that community members were building custom agents with publicly available AI tools, and we wanted to give them something better — a secure, no-cost environment designed for Yale’s needs.

When you develop a custom agent in the Clarity platform, your data stays completely private and isn’t used to train AI systems. The platform offers six models, from GPT to Gemini, so you can pick the one that works best for your specific task. Additionally, we ensure that your deleted chats are purged after 30 days.

This level of privacy and security means you can work with low-, medium-, and most high-risk data safely. This trusted environment promotes more creative and ambitious applications across Yale, rather than having people work around the limitations of public tools.

Does someone need to be a technical expert to build one?

Not at all. If you can write a prompt, you can build an agent.

Anyone with a Yale NetID, including staff, faculty, and students (with a faculty sponsor), can request up to five Clarity Custom Agents. While it’s self-service, our team is here to help you figure out which approach works best for your needs.

Custom agents are especially helpful when you find yourself typing the same prompts or uploading the same documents repeatedly. The agent retains that context and expertise, so you don’t have to re-explain your needs every time.

How are people at Yale using custom agents?

The variety of agents that were developed during the pilot was pretty interesting. Take our own ITS Navigator, for instance, which is a custom agent that helps my peers instantly find information, such as how to book a conference room in our building. Instead of searching Google or calling the Business Office, they get an answer immediately.

In academic settings, School of Medicine students use agents as study partners to quiz them with sample questions and emulate exam conditions. Meanwhile, data science faculty are teaching students to use AI for everything from analyzing datasets to evaluating research pitches.

On the operational side, staff are building agents that onboard new employees, format documents more consistently, and store searchable departmental knowledge. Essentially, people are turning their most repetitive tasks into conversations with a knowledgeable assistant.

How will custom agents change the way people work at Yale?

Custom agents are already making information easily accessible, but we’re just scratching the surface. Right now, they’re replacing Google and document searches with instant, personalized responses. As these agents become more interconnected and autonomous, they will help people get more done and tackle projects that were previously too time-consuming or challenging.

The pilot showed us that when people have the right tools, they innovate in ways we never expected. Our community is already reimagining how AI can enhance teaching, research, and daily work. I’m excited to see what they build next.

What’s the process for getting started?

To learn more about building a custom agent, including information about feasibility, technical details, and support, visit the AI at Yale website. There, you’ll fill out a form with details such as who needs access, and what type of data you’ll use. Our AI Platforms team reviews your request and provides recommendations. Depending on your use case, you’ll typically get access within a few days, along with self-serve resources to help build your first agent.