Citywire Seeks to Deepen Audience Insights via AI-Powered ‘Ask Citywire’
UK-based media company Citywire launched a new AI-powered tool, “Ask Citywire,” to drive better insights on its readership, “crucial” data points for advertisers.
“We can see what questions people are asking and that just gives us another layer of knowledge about what our readers are interested in,” Richard Lander, director at Citywire, told AMO.
Citywire is a publisher that serves the asset management and wealth industries with titles such as Funds Insider and Wealth Manager, as well as proprietary data tools and over 110 live events. The London Stock Exchange Group (LSEG) is among the backers of the privately-owned company, which celebrated its 25th anniversary last year.
Ask Citywire, which sits at the bottom of articles, allows readers to click on related questions or ask their own questions about an article. It is currently only operational on Investment Trust Insider, but there’s a plan to roll it out across Citywire’s other publications depending upon reader feedback.
The decision to build Ask Citywire came from a chance meeting between Lander and Lucky Gunasekara, the CEO of Miso Technologies, at the FIPP publishing conference in Cascais, Portugal, last year.
“One of those things where you wander into the wrong stream [at a conference] and then you say, well, this guy’s interesting, so I’ll listen to him,” Lander said. The two got to talking and in just 10 months, Ask Citywire went from a concept to a live tool on the site.
Miso, a San Francisco-based startup, sells AI-driven search tools for publishers via a “simple flat license fee,” Gunasekara told AMO. The technology is citation driven and powered by private LLMs that run on top of the publisher’s content catalog meaning all responses can be linked back, with citations, to the publisher’s own content. PCWorld, O’Reilly and Nursing Times are listed as existing partners. Publishers are seeing two times registrations, and subscriptions up 10% to 15%, Gunasekara said.
“We don’t charge by usage—our usage levels are usually much higher than teams suspect, and that can freak out the finance team and force us to suddenly cut access or throttle usage, which doesn’t make sense,” Gunasekara said. “And we are pretty reasonably priced these days, with plans for a free tier soon. My goal with this is to always put money into the newsroom, not take money out of it, and keep growing our federated decentralized network of partnerships around the world.”
Safe From Hallucinations
Citywire’s tool is designed so that questions asked on Investment Trust Insider will only get answers from that publication and not from other Citywire sites. It’s fairly safe from hallucinations, Lander said. Companies—especially those that are highly regulated such as the investment industry—are increasingly building internal LLMs due to fears around hallucinations and the misuse of internal data. Citywire shares a similar “skeptical view.”
“We’re not licensing to anybody,” he said. “We keep a very tight lid on preventing these LLMs from crawling over us and picking up content. That’s our attitude, so by the same token, we want to keep everything inside our own particular baseball stadium.”
The AI tool has a sponsorship avenue which could contribute towards Citywire’s revenues—similar to how Citywire’s search tool has been sponsored across the various sites. Baillie Gifford, a UK asset manager which oversees 12 investment trusts, currently sponsors Ask Citywire.
“This is very much dipping our toe in the water, getting it technically proven—which I think we have done—seeing what type of response we get and then we’re going to look at, ‘Can we put this on our real 100% professional sites of which we’ve got many around the world.’”
Travel publication Skift rolled out a similar tool two years ago. Last year, the publication told AMO that the tool had received a total of over 23,000 questions over the course of about six months.
Ask Citywire’s model is only at version 1.0, Lander said, which Gunasekara echoed, noting that Miso will layer more into the upcoming upgrade of the Answers engine next month. Lander sees a future where the tool will start delivering more and more content on demand to readers, leveraging not only written articles but also Citywire’s treasure trove of data such as portfolio manager rankings and performance stats.
“I’m also keen to start integrating their data visualizations of trusts into the Answers 3.0, and enable the new 3.0’s capabilities for being able to directly prompt things like deep research reports, analyses, timelines and things you can really see ChatGPT has started to nail recently,” Gunasekara said.
“I think the bar here isn’t helping Citywire leapfrog over their peers with AI, but leapfrogging ChatGPT with the bundle of high quality professional content + high quality professionalized AI.”
Update: Added comment from Gunasekara in the seventh paragraph.
