Medical Dialogues: Personalising news and surveys for India's doctors with AI

Project name: MedSense AI

Newsroom: Medical Dialogues (India)

Solution: RAG-powered content personalisation and survey engine matching medical specialists with relevant articles and surveys

Organisation staff size: 51 - 100

Tech stack: In-house open-source LLMs, RAG architecture, vector databases, Blinkcms AI integration

Impact: Created the foundation for a non-advertising B2B revenue stream via incentivised doctor surveys while boosting reader engagement

Core lesson: Solve editorial engagement and commercial sustainability concurrently by tailoring content to niche audience specialities


Medical Dialogues, an Indian news portal for doctors and healthcare professionals, set out to tackle two challenges with a single tool. MedSense AI uses AI to match each reader with the articles and paid surveys most relevant to their speciality. The aim is to keep readers engaged for longer and to open a new revenue stream beyond advertising.

The problem: One-size-fits-all news for a specialised audience

Medical Dialogues was founded in 2015 to promote medical journalism in India, serving doctors and healthcare professionals. Its audience, however, is far from uniform. “Healthcare professionals are of different categories,” explains Meghna Singhania, Founder and Editor-in-Chief. “There are specialists, there are medical students, there are nurses, there are generalist doctors.”

Each of these readers wants updates relevant to their own area of expertise. “If I’m a cardiologist, I want to know what’s happening in the field of cardiology. I’m not that interested in what’s happening in the field of gynaecology,” Singhania explains. Relevance serves the publication too: readers who find what they need stay longer and return more often.

The second challenge was financial. According to Singhania, programmatic advertising had been the publication’s main source of revenue, but that income has been falling as AI reduces Medical Dialogues’ traffic arriving from search. One bright spot has been first-party advertising, from pharmaceutical companies promoting products to gynaecologists to carmakers offering doctors special discounts. “Anybody who wants to reach out to this targeted community of doctors is coming to us,” Singhania says. Still, she adds, “everybody is trying to search for a revenue model beyond advertisements.”

At the same time, industry partners and researchers kept asking to survey the platform’s audience of around 700,000 registered doctors and were ready to pay for it. The team saw one answer to both problems: the right articles and surveys for the right specialist.

Building the solution: MedSense AI

MedSense AI is an AI-powered content personalisation and audience intelligence initiative developed through the JournalismAI Innovation Challenge. Over nine months, the team worked on two fronts.

The first was a survey module built from scratch. The team began by asking industry partners how they run surveys, and found most still do so by hand. “There’s actually a printout that they take, and they sit with the doctor and get the survey filled,” Singhania explains.

With existing tools falling short of partners’ needs, the team built a back end supporting multiple-choice, single-choice and open-ended questions; a front end that clients can white-label; and a points system that rewards doctors with vouchers for completed surveys.

The second front was the AI itself. Doctors state their speciality when they register, but that does not tell the platform what to show them as they read. To bridge the gap, Vinoth Poovalingam, CEO of technology partner Blinkcms AI, and his team implemented a RAG system. It maps every article into vector space, allowing it to understand semantically whether a story concerns cardiology or gynaecology, for instance.

“If someone is reading a cardiology article, we’ll be able to show them the next cardiology-related article,” Poovalingam says. “Second thing is, if we have a survey which is targeted to cardiologists, we’ll be able to show that.” After experimenting with OpenAI, the team settled on open-source models, hosted in-house and customised for Medical Dialogues.

The biggest hurdle was the lack of a blueprint. “We’ve not seen this kind of a solution before with any other newspaper or any other industry partner,” Singhania says. The budget also shaped the scope: a costly analysis feature was set aside for a later stage. The tool is now live and undergoing full testing.

The opportunities: Engagement and a new revenue model

“I’m realising that my outside traffic is going down, so I need that engagement,” Singhania says. The goal is for a visitor to read five pages rather than two. “If I’m able to run a monetising model both for me and my readers, I’m pretty sure they’re going to be engaged with my platform.”

The next step is a pilot, offering the clients consulted at the start short and complimentary surveys of around 20 respondents in exchange for feedback, before a full launch. In the longer term, the team plans to add analysis features, creating an end-to-end platform for anyone wanting to survey doctors. “It’s going to help medical research a lot in India,” Singhania says.

The model could also travel. Poovalingam notes that few Indian publishers have invested in AI beyond off-the-shelf tools, and his firm works with other media companies that could follow suit. Singhania sees potential for any outlet with a niche audience.

Lessons for newsrooms

Medical Dialogues’ experience offers several lessons for publishers serving specialist audiences:

  • Start with the problem, not the technology: Talk to the partners and readers you serve before you build. Their needs shape the product and confirm there is a market.

  • Let one tool serve two goals: Look for AI projects that solve an editorial and a commercial problem at once. Relevance drives engagement, and engagement makes new revenue models possible.

  • Narrow the scope, stay flexible: Build the smallest version that tests the core idea, set aside costly extras and keep the design open to changing client needs.

  • Personalise responsibly: Personalisation is not purely a technical problem. It demands attention to editorial integrity, transparency, trust and privacy. Relevance should aid discovery, as opposed to creating information silos.

Explore case studies from the 2025 cohort

The 2025 Innovation Challenge Report, Prompting Sustainability, collects case studies from all 12 grantees and reveals their journeys, challenges, and opportunities.

Read about their experiences, discover best practices, and find inspiration in their innovative approaches to integrating AI into journalism.


The JournalismAI Innovation Challenge, supported by the Google News Initiative, is organised by the JournalismAI team at Polis – the journalism think-tank at the London School of Economics and Political Science, and it is powered by the Google News Initiative.