AI transformation for digital publishers [Part 1] - Reflecting on success stories so far

In the past year, publishers have launched various AI-driven products to enhance their offerings and engage audiences more effectively. From AI-driven games to subscriber-only conversational tools, some publishers are taking the AI opportunity with both hands and the results are more than encouraging! Let's dive in

In this edition, I analyse the customer-facing AI innovations that publishers have launched. I double-click on

• The rationale publishers gave in their press releases
• Details about how it was technically executed
• The impact that was created - data from interviews and case studies

Towards the end, I also share how you can ensure these innovations do not create a negative impact on your reputation and how you can get a human in the loop approach to ensure there are no hiccups!

Before we begin, I want to share a quote from Lindsey Jayne , Chief Product Officer at FT that she said in an interview while reflecting on FT's Gen AI initiatives-

Generative AI is an example of the ‘chicken or egg’ dilemma between what a user needs and what technology can do. You should always start with a user need and build something to meet it, but sometimes technology comes around that can do things that weren't possible before, and you have to start testing the limits.

I always recommend that any media owner think of AI as a means to an end. If you cannot pinpoint what problem your AI initiatives can solve for you or what KPI will impact, you are not ready for AI adoption. Let's learn together from publishers that have been able to successfully launch customer-facing AI products.

The most launched Customer-facing AI products have been conversational chatbots, summary generators, and polls [As per an analysis by Bridged Media team where they analysed 300 publishers in the UK, Spain, Germany, the US and India].

Conversational Chatbots

Examples:

• Ask FT by Financial Times : This AI chatbot, powered by Anthropic's Claude model, is designed to answer subscriber questions using the Financial Times' extensive archive of articles.
• Adelaide by Forbes : Developed using Google's Enterprise Search, this tool allows users to search Forbes' content in a conversational manner, initially using content from the past 12 months. [I remember using this tool but I can't find it anymore. Maybe it is geo-targeted now or it could be a ripple effect from the Forbes MFA scandal]
• ZEIT ONLINE : One of Germany's biggest publishers launched an AI-powered chatbot in August 2024 to help users find articles, answer questions, and provide summaries, enhancing content accessibility and user engagement.

Hypothesis and Benefits:

• Summary of why they said they built conversational chatbots: Publishers aim to leverage their unique content archives to create AI tools that provide users with personalized and contextually relevant information, enhancing engagement and loyalty.
• What KPI got impacted: In a survey for FT Ask, 75% of people said they found it helpful [source - FT Strategies]. A study published by Ume? universitet found that conversational experiences increase loyalty.

Development:

• Ask FT: Developed in collaboration with Anthropic, utilizing their Claude large language model.
• Adelaide: Built using Google's Enterprise Search, allowing Forbes to exclusively use its own content.
• Zeit Online: Built with Azure and Deepset Cloud
• Off-the-shelf products - Check out bridged.media that have helped 20+ publishers launch such experiences with no tech sprints

Article Summaries

Examples

• Clarín (summaries branded as Walter AI): This Argentine newspaper has implemented AI to generate content snippets to provide readers with quick and accessible summaries of news articles.
• Schibsted : This media group has integrated AI-generated summaries across its newsrooms. Initially developed as a hack-day experiment, these summaries have become a standard feature.
• nuevecuatro : This regional Spanish newspaper has launched reading assistants that show summaries, key facts, quotes and other articles based on user interaction.

Hypothesis and Benefits:

• Summary of why they said they built Article Summaries: Summary generators are developed to cater to the fast-paced consumption habits of modern readers who prefer quick access to essential information.
• What KPI got impacted: Schibsted reports that articles with AI-generated summaries have open rates between 20% and 30%, with particularly strong engagement among younger readers. Contrary to initial concerns, the addition of summaries has increased the time spent reading articles and did not impact the scroll depth

Development:

• Clarín and Schibsted: Developed in-house using generative AI models that can process and summarize large datasets. They then use their frontend and backend teams to create supporting systems to show summaries and collect user feedback.
• NCU, Enter.co, Nacio Digital, among 30+ publishers have used out-of-the-box Reading assistants from bridged.media

Interactive polls

Examples

• TIME : Launched AI generated Quiz experiences that the audience could interact with and get AI-generated feedback
• Nació : Launched opinion polls on the Living Well website that asked the opinion of people on their articles in the form of slider polls
• Farmers Weekly : An AI-generated quiz from the latest articles published with links to those articles once the user responds

Hypothesis and Benefits:

• Summary of why they said they built Interactive Polls: These tools are designed to engage users actively and show them more contextual content. They can also be used to collect data on reader preferences and opinions, which can inform content strategy or create user profiles
• What KPI got impacted: Increased interactions, More customer profiling, Higher pageviews

Development:

• Time launched it through its internal teams working with ChatGPT
• Farmers Weekly used an external vendor called Direqt, while Nacio used Bridged Media

Key considerations before launching customer-facing ai products

Address these 3 critical factors to ensure the success and reliability

Establish Guardrails

• Use relevant datasets to train the AI, ensuring it understands your audience’s needs and can maintain your brand’s voice.
• Proper fine-tuning helps prevent issues like biased outputs or inappropriate content, ensuring the AI behaves responsibly.
• Beyond fine-tuning, implementing verification processes for AI-generated outputs is crucial. This can include using secondary AI models to cross-check and validate content before it reaches customers.

Keep a Human in the Loop

• While AI can automate and enhance many tasks, human oversight is essential to ensure accuracy and relevance.
• This combination of AI efficiency and human discernment helps maintain a trustworthy and reliable customer experience. Below is a Chrome plugin we give to any editor using Bridged′s AI agents --> To approve or update any AI-generated experiences on a page and check feedback from visitors
• Creating such tools will give peace of mind to all stakeholders

Focus on Transparent and Ethical AI Use

• Clearly communicate to your customers when and how AI is being used in your products.
• All experiences in the market clearly have labels on how this is a Gen AI-developed experience.
• Most experiences also mention there is a non-zero likelihood of incorrect information. Eg: Zeit Online mentioned, “The likelihood that our AI is making up information is low, as it only uses our articles as a source.”

If you are looking to unlock growth with AI experiences, drop me a note. We have done 500 AI experiments with publishers globally and I would love to share learnings specifically to your use case.

About Me

I am an experienced entrepreneur who has worked in Media and Tech consulting for many years. Apart from being a full-time parent, I am also leading a project that aims to make AI more accessible to level the playing field between tech vendors and digital publishers in the Media ecosystem. Bridged’s out-of-the-box agents eliminate the need for extensive data processing or dedicated AI resources, making AI adoption accessible and efficient.