In recent years, AI for publishing has ceased to be a simple “experiment” and has become a daily presence in the publishing world. The GenAI Communication Industry Report confirms that 80% of media professionals use generative AI tools to write, edit, and even produce images and videos.
But the real revolution is not limited to production: the way readers discover and consume content is also changing, with major implications for SEO, information quality, and editorial models.
In this scenario emerges the Librarian project, the “digital librarian” chatbot created by CMI Customer Management Insights together with the Crafter AI platform, and presented in the webinar with editor-in-chief Letizia Olivari.
AI for Publishing: Content Management Artificial Intelligence
In recent years, AI for publishing has ceased to be a simple “experiment” and has become a daily presence in the publishing world. The GenAI Communication Industry Report confirms that 80% of media professionals use generative AI tools to write, edit, and even produce images and videos.
But the real revolution is not limited to production: the way readers discover and consume content is also changing, with major implications for SEO, information quality, and editorial models.
In this scenario emerges the Librarian project, the “digital librarian” chatbot created by CMI Customer Management Insights together with the Crafter AI platform, and presented in the webinar with editor-in-chief Letizia Olivari.
Content Management Artificial Intelligence: What Is Really Changing
According to Letizia Olivari, the impact of AI in newsrooms is now undeniable, even though Italy still shows some delay in adopting more advanced applications.
Among the most common uses of content management artificial intelligence:
- Automatic transcription of interviews, once a highly time-consuming task for journalists
- Subtitle generation for live streams and videos
- Summaries of existing content, optimized for web and social formats
- Multilingual translation to expand international reach
- Automated fact-checking and dataset analysis to support investigative journalism
A study by the Italian Order of Journalists conducted with LUMSA University confirms that, despite growing familiarity with AI tools, adoption is still low for activities such as autonomous writing or image generation, while translation and transcription are now well-established practices.
As Olivari emphasizes, technology does not replace journalists: source verification, ethics, and editorial responsibility remain essential. AI is a powerful tool — but one that must be governed.
Agreements Between Publishers and OpenAI: Opportunity or Risk?
Among the emerging scenarios enabled by content management artificial intelligence, one that raises important questions is the choice of many publishing groups — in Italy and abroad — to sign agreements with OpenAI, allowing the use of their archives to train generative models.
According to Olivari, this choice has both strengths and risks.
Risks
- Source homogenization: if all models read the same content, information pluralism shrinks.
- Disintermediation: if the answer comes from the model instead of the original article, journalism’s role weakens.
- Technological dependence on external and often opaque systems.
Opportunities
- Defining more ethical, regulated training models
- Creating new services for publishers based on controlled AI use
It remains an open debate: “Technology is here to stay,” Olivari notes. “We must work on it, not endure it.”
The Audience Is Changing Too: Consumption Becomes Immediate and Conversational
Answer engines (Google AI Overview, ChatGPT, Perplexity) are now the first gateway to information. Before even clicking any link, the user receives a synthetic — and often sufficient — answer.
This phenomenon creates three main effects:
- Drop in traffic to original pages: if the summary suffices, readers don’t click through
- Reduced attention time: immediacy is increasingly expected
- Harder-to-interpret analytics: how can publishers understand what was truly read if consumption happens “upstream”?
These changes require publishers to rethink writing practices in light of AI for content management, making editorial outputs:
- more modular
- more navigable
- more aligned with a question–answer structure
- more readable for AI engines and answer engines
Innovative Use Cases of Content Management Artificial Intelligence
Internationally, several noteworthy use cases are emerging:
- The Washington Post: introduced a dedicated AI section that helps readers navigate news, answer questions, and find relevant content
- Corriere della Sera: uses AI for article audio-reading and, for subscribers, personalized search and recommendations
- Southeast Asian outlets: experimenting with automated fact-checking on complex datasets
A case particularly similar to CMI’s project is Gartner’s AI assistant, designed to help users explore its research library.
The CMI Project: Librarian, the Digital Librarian Chatbot
Back in 2021, CMI had already created Library, a knowledge base of about 500 Q&As on customer experience — a valuable resource but difficult to navigate with traditional search tools.
From this foundation came the idea to use content management artificial intelligence: Librarian, the digital librarian chatbot created with Crafter AI.
How Librarian Works
- It is trained on CMI’s certified content: knowledge bases, articles, interviews, research, and white papers
- It analyzes user questions, identifies the topic, and retrieves the most relevant content in seconds
- It creates intelligent connections, suggesting related readings
- Its knowledge updates dynamically as CMI publishes new material
- It offers micro-learning paths generated from user queries
- It helps the editorial team understand which topics interest readers most
Importantly, Librarian maintains a style consistent with CMI’s editorial identity: it is not a general-purpose AI but a specialized assistant built on sector-specific, verified knowledge.
Why Not Just Use ChatGPT or Gemini? The Difference Between General AI and Expert AI
Olivari answers one of the most common questions:
“Couldn’t ChatGPT provide the same information?”
The answer is no, and for several reasons.
ChatGPT delivers probabilistic answers
- It draws on a huge corpus, but not always updated
- It doesn’t guarantee alignment with the language of customer experience
- It doesn’t reflect CMI’s editorial approach
- It may invent examples or case studies not present in the organization’s corpus
Librarian, instead, is an expert
- It answers only from CMI’s verified content
- It maintains the editorial tone
- It provides access to recent research not yet absorbed by generalist models
- It suggests paths, correlations, and contextual links
- It avoids hallucinations
In short, it’s the difference between a generic encyclopedia and a librarian who knows every detail of your archives.
Librarian is a pioneering case in Italy and, according to Olivari, can inspire:
- schools
- universities
- publishers
- companies producing specialized content
Its potential is enormous: AI unlocks new ways to explore knowledge, enhancing human expertise and freeing time from repetitive tasks.
Conclusion
While the fear that AI may replace human work persists, Letizia Olivari argues the opposite: content management artificial intelligence actually highlights the value of human skills. In many sectors, work has become flattened around routine tasks, and the arrival of AI forces us to “look up,” developing critical, interpretive, and analytical abilities that no algorithm can replicate.The Librarian project proves that when AI is governed, trained on verified content, and integrated with a strong editorial vision, it becomes a strategic ally. It improves content discovery, increases engagement, enhances archives and knowledge capital, supports editorial teams, and preserves the identity of those who create content.
It is not a substitute for journalists but an accelerator of quality and knowledge.