At The Quint, a digital-first Indian news platform, an AI-powered tool called NewsEasy is helping readers engage with long-form journalism in ways that match their preferences without altering the core reporting. The initiative emerged from a recognition that although audiences showed interest in in-depth stories, their actual engagement—particularly on mobile and social platforms—was not matching that intent. This gap prompted the newsroom to explore how presentation could be adapted to serve different reader needs, especially among younger audiences who indicated a preference for varied formats over traditional linear reading.
The Quint, founded in 2014, has built its reputation on explanatory journalism, mobile-friendly storytelling, and fact-checking through its WebQoof initiative. As audience behaviors evolved under the influence of social media, search algorithms, and AI-driven content consumption, the platform sought support through the 2025 edition of the Newsroom AI Catalyst programme—a collaboration between WAN-IFRA and OpenAI designed to assist news publishers in developing strategic AI initiatives. It was within this framework that The Quint began developing NewsEasy, not as a replacement for journalism, but as a supplemental layer to enhance accessibility, and interactivity.
The core problem identified by The Quint’s product team was a mismatch between high audience intent and low sustained engagement. Tarun Jain, Product Head at The Quint, noted that while page views for long-form exclusives were rising, average time spent on articles remained below desired levels. Scroll depth data revealed that many users were not progressing far into stories, signaling an opportunity to deepen interaction. Jain explained that the issue was not with the journalism itself but with how it was being presented across different audience cohorts, particularly younger users who were interested in topics but unwilling to commit to lengthy formats.
This insight led to the development of NewsEasy—a sticky, embedded widget that appears within articles and offers three distinct, AI-generated formats: a concise “Article-in-Brief” summary, five key takeaways, and a Q&A-style breakdown. According to Jain, readers can choose how deeply they wish to engage: those in a hurry might read only the brief, others seeking more structure might review the takeaways, and those aiming to understand complex issues can apply the Q&A section. The design ensures the widget is optional and non-intrusive, allowing readers to ignore it entirely if they prefer the full article.
Crucially, all content generated by NewsEasy is strictly grounded in the original article. Abhilash Mallick, Editor of WebQoof at The Quint, emphasized that the system is designed to avoid introducing new information or hallucinating outputs. Significant effort went into prompt engineering and guardrails to ensure factual fidelity. Each format—summary, takeaways, and Q&A—is generated separately using tailored prompts, but all draw exclusively from the source text. This approach allows the AI layer to adapt presentation without compromising journalistic integrity.
Implementation was carefully structured to avoid disrupting editorial workflows. NewsEasy is applied only after a story has been written and edited, functioning as an additive layer rather than a replacement for any editorial step. Editors decide whether to activate the widget, typically for longer pieces, explainers, or stories where higher drop-off is anticipated. Once generated, outputs undergo human review, especially for sensitive topics, reinforcing a human-in-the-loop model where journalists retain final oversight.
The development process was iterative. Initial prototypes emerged from the product team, followed by close collaboration with editorial, engineering, and audience engagement teams. Prompts were refined through repeated testing, and a feedback loop was integrated to flag inconsistencies and improve accuracy over time. On the technical side, the pipeline routes content from the CMS through an automation engine to a custom backend before rendering on the website. Version control is maintained via prompt files, enabling quick adjustments as needed.
Early results from the pilot phase show promising engagement signals. Jain reported improvements in scroll depth, particularly on longer stories, with targeted users staying engaged for longer periods. In several cases, overall time spent on articles also increased, suggesting more sustained interaction with the content. The team continues to test widget placement and format effectiveness as part of a second iteration, aiming to optimize both user experience and editorial utility.
One key lesson learned during development was the limitations of a single-format approach. Initially, the team considered offering only a summary, but quickly realized that different readers have different needs. The shift to multiple formats—summary, takeaways, and Q&A—made the tool more versatile and engaging. At the same time, maintaining tonal and stylistic consistency required a structured response. The team created a centralized prompt framework aligned with The Quint’s editorial voice: clear, factual, and non-sensational—while still allowing desk-level adjustments for nuance.
Operational challenges also surfaced, particularly around API costs. Without optimization, token usage and associated expenses rose rapidly. To address this, the team implemented batching and caching strategies that reduced token consumption by 38 percent, a significant saving that made the tool more sustainable at scale. Another important insight concerned transparency: while early metrics showed improved engagement, the team recognized the need for clearer disclosure about AI involvement to maintain reader trust. This has become a focus in ongoing refinements.
Looking ahead, The Quint plans to scale and refine NewsEasy. Future phases may include new formats such as timelines, enhanced metadata, podcast integrations, and reader feedback mechanisms. The team is also exploring language support to broaden accessibility and more advanced query handling to enable deeper interactivity. Infrastructure improvements, like a cost-optimized embedding store to pre-cache high-traffic stories, are under consideration to improve efficiency.
Despite these expansions, the central philosophy remains unchanged: adapt how stories are presented, not what they say. As Jain stated, the goal is not to alter the reporting but to “build on our efforts to distribute content in personalised formats to enhance interactivity and accessibility” so that each piece can serve wider audiences. This approach reflects a broader trend in digital journalism—using AI not to generate news, but to improve how existing journalism is experienced by diverse readers in a fragmented media landscape.