Exploring AI Summaries in Scholarly Publishing: A Conversation with Bristol University Press and KGL PubFactory

by Adrian Stanley

From left to right: Victoria Pittman, José Fossi, and Alison Shaw at the London Book Fair 2026

As AI continues to reshape digital discovery, scholarly publishers are beginning to explore how these technologies can responsibly support research accessibility and engagement. In this conversation, I spoke with Victoria Pittman and José Fossi about their recent AI summary pilot for selected Policy Press titles on Bristol University Press Digital.

AS: Victoria, what initially motivated Bristol University Press to participate in this pilot? 

VP: For us, this was an exciting opportunity to experiment with emerging technologies while remaining focused on our core mission: improving the visibility and accessibility of impactful research. Policy Press has always engaged audiences beyond academia, so the idea of structured chapter summaries was particularly compelling. 

We wanted to better understand whether AI-generated summaries could improve discoverability, support navigation, and help broaden engagement with our authors’ work. Just as importantly, we wanted to gather real user feedback and meaningful usage data to help inform future decisions. 

AS: José, from KGL’s perspective, what made this pilot technically different from simply generating AI summaries with a prompt? 

JF: One of the biggest misconceptions around AI is that you can simply feed content into a model and immediately trust the output. Scholarly publishing requires a much more rigorous approach. 

For this pilot, we built a framework specifically designed for academic content. A key part of that was our multi-agent methodology, where different AI agents evaluate summaries against specific criteria such as factual accuracy, readability, and structural quality. We also benchmarked outputs against human-reviewed reference content to improve consistency and reliability. 

The goal was not just speed, but trustworthiness. 

AS: Victoria, how important was publisher control and transparency throughout the process? 

VP: Extremely important. We wanted the summaries to feel aligned with the Policy Press brand and audience expectations, rather than sounding generic or overly automated. KGL was very collaborative in refining tone, terminology, and structure based on our feedback. 

Transparency was equally critical. We felt strongly that users should clearly understand when content was AI-generated, and we also saw real value in implementing summaries at the chapter level rather than only at the title level. That granularity helps readers navigate content more effectively. 

AS: José, what stood out most to you during the collaboration? 

JF: Honestly, the pace of learning, executing from our teams. Because the Bristol University Press team was highly engaged and collaborative, we were able to move very quickly from concept to implementation. 

That kind of partnership is essential when experimenting with emerging technologies. It allowed us to iterate rapidly, incorporate feedback continuously, and explore different presentation and workflow approaches together. 

AS: Have you seen any early indicators around user engagement? 

JF: It’s still early, and we’re continuing to analyze the data, but yes, the initial signals are encouraging. We’re already seeing indications of increased engagement, including 2.5X longer average session times as users interact with summaries to navigate more deeply into the research itself. 

That’s important because the summaries are not intended to replace the content — they’re intended to guide readers into it. 

AS: Victoria, how have authors responded to the initiative? 

VP: The response has generally been thoughtful and constructive. Many authors recognize that summaries can act as a valuable discovery and marketing tool, especially in an increasingly AI-driven information landscape. 

For us, this pilot has also opened important conversations around how publishers can responsibly use AI to support authors while maintaining trust, transparency, and scholarly integrity. 

AS: Finally, what comes next? 

JF: From KGL’s perspective, this is only the beginning. We see AI-enhanced discovery as an evolving capability that will continue to mature through collaboration with publishers. KGL already has and will continue to invest in its technology & AI infrastructure / capabilities. I invite you to follow what we are doing around the KGL AI Gateway.

VP: And for Bristol University Press, the next step is learning from the pilot — combining community feedback, editorial input, and usage analytics to determine how these tools can best support readers and authors moving forward. 

What’s been most encouraging is that this work demonstrates AI can be implemented thoughtfully, transparently, and in ways that genuinely enhance scholarly communication. 

KnowledgeWorks Global Ltd. (KGL) is the industry leader in editorial, production, online hosting, and transformative services for every stage of the content lifecycle. We are your source for society services, market analysis, intelligent automation, digital delivery, and more. Email us at info@kwglobal.com.