How AI Is Changing Research Discovery: An Interview with Kamran Kardan

1. What are the biggest challenges libraries and research organizations are facing today when it comes to research discovery and access?

I think the biggest challenge is that there is more information available than ever before, but finding the right information is becoming more and more difficult.

If you look at the research ecosystem, you’ve got millions and millions of articles, books, journals, databases, different platforms, different publishers, and different discovery systems. The challenge is being able to discover the right information at the right time and then being able to actually work with it.

And then you have the access challenge. There is still a huge amount of scholarly information behind paywalls and subscription models. Of course, there is a value chain behind publishing, and publishers provide an important service in the research ecosystem. But at the same time, we have a global challenge around access to knowledge.

And from a library perspective, there’s another challenge, which is that libraries are investing significantly in content and resources, but if researchers cannot easily discover those resources, then you’re not necessarily getting the full value from that investment.

So for us, discoverability is very important.

So we need to look at how we can make research more accessible, more discoverable, and easier to work with, while maintaining the quality and trust that scholarly communication requires.

And this is where I think AI gives us a very interesting opportunity.

2. What impact can better discovery have on the visibility and usage of a library’s existing resources?

I think this is one of the biggest opportunities for libraries.

Libraries can have fantastic collections. They can subscribe to some of the best journals and databases in the world, but if researchers don’t discover those resources, then the value of those investments is limited.

So when we talk about improving discovery, we’re not only talking about making the researcher’s life easier. We’re also talking about helping institutions get more value from the resources they already provide.

If I have a researcher who comes into a library discovery system and searches for something using a very specific keyword, there is a possibility that they won’t find the information they’re actually looking for because they don’t know the terminology or they don’t know which database contains it.

AI can help bridge that gap.

It can understand the intent behind the question and help connect the researcher with relevant resources across the institution’s existing ecosystem.

And that can have a direct impact on visibility and usage.

Ultimately, libraries want their resources to be used. They want researchers and students to discover them, read them, cite them, and build new research on top of them.

So better discovery is really about making the value of the library more visible to the people it serves.

3. How does Zendy AI Discovery complement the discovery systems and infrastructure that libraries already have in place?

One thing that was very important for us when we were developing Zendy AI Discovery was that we didn’t want to come in and say, “Forget everything you’ve already invested in and what you have built so far. Let’s replace all your existing workflows and systems.”

That’s not realistic, and actually it’s not necessary.

Libraries have invested years in building their infrastructure, their discovery services, their link resolvers, their databases, their collections, and their workflows. There is a lot of value in that ecosystem.

So what we wanted to do was add an AI layer on top of that that acts as a research partner.

Zendy AI Discovery is designed to work in your existing environment and help researchers interact with scholarly resources in a more natural way. Instead of simply relying on keyword matching, you can start thinking about the intent behind the researcher’s question and the context of what they’re looking for.

And importantly, the institution remains in control and becomes that gateway to an AI tool that everyone can trust.

So we’re not trying to take away the role of the library or the existing infrastructure. We’re trying to make that infrastructure more beneficial for researcher and the library. 

If you think about it, the library has already done a lot of the hard work. They’ve selected the resources, they’ve subscribed to the databases, they’ve established the systems. What we’re trying to do is make it easier for the researcher to actually get to those resources and make use of them.

4. Trust is a major concern when institutions consider AI. How does Zendy approach accuracy, citations, and the use of trusted scholarly sources?

I think trust is probably one of the most important questions when we talk about AI in research.

Because if you’re asking an AI system what restaurant should I go to, maybe you get a bad recommendation. It’s not the end of the world.

But if you’re a researcher and you’re using AI to understand scientific literature, or you’re working on a medical question, or you’re making a policy decision, then the consequences can be much more significant.

And we saw very early on that hallucination was one of the biggest challenges with large language models. You can get a very convincing answer, but the question is: where did that answer come from?

For us, traceability is extremely important.

We want researchers to be able to understand where information is coming from and to have references that they can actually follow and verify. That’s why our approach is based around trusted scholarly content, appropriate licensing, and citations. We believe the content used in AI-supported research needs to come from sources that institutions and researchers can trust, with the appropriate rights and permissions in place for how that content is accessed and used.

We also believe that AI should be a tool, not a replacement for human judgment. A researcher still needs to evaluate the information. A librarian still has an important role. An academic still needs to look at the original source and make a judgment about its relevance and quality.

The technology should help people get there faster and make the process easier, but the human element remains fundamental.

And this is also why we have been very transparent about our AI principles, our approach to scholarly content, and the licensing considerations involved in working with that content. When you’re dealing with research, people need to know what is happening with the information, where the answers are coming from, and whether the content is being used within the appropriate rights and permissions. Licensing is an important part of the wider conversation around AI and scholarly research, and it’s something we believe needs careful consideration as this space continues to develop.

5. What advice would you give to library leaders who are interested in AI but are still unsure where to start?

I would say start with the challenge, not with the technology.

There’s a lot of excitement around AI, and every organization is asking, “What can we do with AI?” But I think the better question is, “What challenge are we trying to solve?”

Look at your researchers. Look at your librarians. Look at the workflows that are taking a lot of time. Look at where people are struggling to discover information or where valuable resources are not being used.

Then ask yourself: can AI help us solve this particular problem?

And I would also say don’t try to do everything at once.

Start with a practical use case. Test it. Talk to your users. Talk to your librarians. See what works and what doesn’t work.

And importantly, look at trust and governance from the beginning. Where does the information come from? What content is being used? How are answers generated? Can users trace the information back to the source? What control does the institution have? These questions are as important as the technology itself.

AI is moving very quickly, but libraries don’t have to rush into every new development. They can take a practical approach and introduce AI where it genuinely improves the research experience.

6. What is the one thing you would want library and research leaders to understand about Zendy’s approach to AI and research discovery?

I would say that our fundamental purpose has always been about making knowledge more accessible and more discoverable.

We started Zendy because we believe that access to research is essential if we’re going to address some of the major challenges that the world is facing.

And when AI came along, we saw a huge opportunity, but also a huge responsibility.

So our approach is very simple in principle: use AI as a tool to help people discover, understand, and work with scholarly information, while keeping trust, transparency, and human judgment at the center.

For institutions, that means we want to work with the ecosystem that already exists.

We don’t want to replace the library. We don’t want to replace the librarian. We don’t want to replace the researcher’s judgment.

We want to help all of them work more effectively with the enormous amount of knowledge that is available today.

Because at the end of the day, technology is only useful if it helps people do something better.

And for us, that means helping researchers discover the right knowledge and improving their research quality, helping institutions get more value from their resources, and ultimately making scholarly information more accessible to the people who need it.