What Happens When the News Starts Answering Back? In Conversation with Miso.ai CEO, Lucky Gunasekara

Lucky Gunasekara’s career doesn’t follow a straight line.

He’s researched intellectual property, studied neurobiology, and built a massive RSS crawler to track global disease outbreaks by pulling in real-time data from hundreds of news sources.

That crawler became a research engine, serving answers from academic papers. It eventually became the foundation for Miso.ai, the answer engine company Lucky now leads.

On May 20th, Lucky joined Twipe CEO Danny Lein for the second edition of our AI Frontrunners in News webinar.

We had questions. Lucky had answers, which was fitting. Below, we highlight the main takeaways from our conversation.

What actually happens when you give users answers instead of articles?

Miso’s earliest partner was O’Reilly Media, the academic publishing giant. They integrated Miso’s answer engine into their platform and found that usage went up.

You see a much higher rate of usage when you move away from high-friction search to very frictionless answers.

Lucky Gunasekara

CEO & Co-Founder, Miso.ai

More importantly, engagement changed. Users didn’t just search for something and get a wall of results. They instantly got exactly what they needed. That changed how they thought about O’Reilly’s entire archive. “The value proposition suddenly shifts from ‘I have all these books on the wall’ to ‘What do you need right now?’”.

This, in turn, drove more B2B sales, individual subscriptions, and, critically, led to higher retention.

But what about the authors?

O’Reilly didn’t keep the benefits for themselves. “They actually worked very closely with the author community. They reworked the royalty system as we rolled out answers.”

To illustrate how, Lucky takes the example of Chip. Chip wrote a bestselling book on AI Engineering. Normally, he’d get royalties when someone buys his book or reads it digitally. But with Miso.ai’s answer engine, users may never open the book. They would just get an answer that draws from it.

So how does Chip get paid?

“We calculate the attribution percentage of Chip’s book in the answer and then we calculate the value of the answer and we pay out a royalty against that.”

And because answers are used far more often than traditional search, those royalties add up. This could be a totally new monetization model.

Could this work for news publishers?

Chip wrote a book. Publishers are writing hundreds of shorter articles, often with a lifespan of hours, not years. Can this model work for them?

It’s not a simple copy-paste. The structure of news is different: The content lifecycle is faster and attribution can get tricky. But the idea of a royalty engine that rewards contribution to machine-generated answers is worth exploring. We’ve interviewed Tollbit CEO and Founder, Toshit Panigrahi, who shared his views on how publishers can make money from their content from bots by paying a toll for each scrape.

Lucky also believes the bigger shift isn’t about search engines or bots: it’s about how information moves. “Instead of having solid, static media like articles or constrained interfaces like search or newsletters, you get to a place where information is more remixed and repackaged directly in line with your interests.” This is the age of liquid content.

Project Sentinel: How AI companies are still getting your data

Let’s say you’re a publisher and you’ve read all the AI copyright lawsuits, updated your robots.txt, and told the bots to stay out.

Great job. Doesn’t matter. Or rather, a lot of bots simply ignore or work around this.

We’re now monitoring about 7,000 publishers and news sites around the world every day […] And we’ve realized that 20% to 50% of traffic is just [bots].

Lucky Gunasekara

CEO & Co-Founder, Miso.ai

With Project Sentinel, Miso.ai is tracking how AI companies access news data. They’ve found that many bots are accessing content illegitimately.

Take Perplexity, for example. Many publishers have explicitly said “don’t crawl us” in their robots.txt files, and yet their answers still show up in the aggregator.

How high? “Back in February, 15% of the time, you would be able to get an article. Today, based on the latest tests we ran, with a fairly straightforward prompt, we get 90% of the time, you could get the article.” In other words, even when publishers explicitly block AI bots, the success rate of retrieving their articles through workaround prompts has jumped from 15% to 90% in under three months.

Even though certain AI company bots may be blocked, other third-party bots aren’t. Welcome to the gray market of scraper proxies and commercial “content partners” you never agreed to.

So what can publishers do?

Lucky lays out a three-part playbook:

1. Update your robots.txt files. (42% of publishers haven’t according to Lucky)
2. Turn on bot-blocking. Yes, it comes at a cost, but compare that to the cost of being scraped.
3. Push for regulation. Industry-wide litigation and lobbying will be needed to clean up the current Wild West.

TL;DR

• Publishers must update robots.txt files, block bots, and push for policy change, but also take control by building their own AI-powered user experiences.
• Answer engines drive higher engagement and retention than traditional search.
• O’Reilly Media pioneered a royalty model for answers that could inspire similar models for news.
• Despite blocking bots, AI companies are still accessing and monetizing publisher content, often via opaque methods.
• Project Sentinel is monitoring how this happens and sounding the alarm.