Wikitree picks Taboola DeeperDive AI engine for reader engagement
Wikitree, the South Korean news platform, has selected DeeperDive, a generative AI answer engine developed by NASDAQ-listed Taboola, to embed conversational search directly within its publisher website. The partnership, announced on 1 July 2026, is intended to lift reader engagement, extend session depth, and open new advertising revenue streams for the publisher.
DeeperDive sits on-site rather than redirecting readers to a third-party search interface. It draws on the publisher's own archive of journalist-produced content to generate answers, surfacing related articles and prompting follow-up questions within the same browsing session. Taboola positions the product as part of its broader "Agentic" roadmap, in which AI agents act on behalf of readers and advertisers across the open web.
The deal
Dong-Ki Lee, chief executive of Wikitree, said the company has spent years building content designed to move "from the moment to conversations," and described the DeeperDive integration as the next step in that editorial strategy. "For the first time, our readers can have conversations with our news directly, discovering more about topics they care about, using an innovative generative AI experience," Lee said.
Adam Singolda, chief executive and founder of Taboola, framed the product as a means for publishers to participate in the generative AI shift on their own terms rather than ceding audience to external AI-powered search products. Taboola did not disclose deal financials, revenue-share terms, or the volume of content indexed under the integration.
On monetisation, Taboola said DeeperDive has the potential to insert high-intent advertising directly into AI-generated results pages, creating what it describes as a new category of contextually targeted ad inventory. No live revenue figures or advertiser names were provided in the release.
Market context
The DeeperDive proposition sits at the intersection of two pressures currently reshaping digital publishing: the steady decline in referral traffic from social platforms, and the growing share of search queries being absorbed by AI-generated summaries on Google, Perplexity, and Microsoft Copilot. Publishers that historically relied on search-engine optimisation for audience growth are reassessing their distribution strategies as zero-click results erode click-through rates.
Several technology vendors are pursuing similar on-site AI experiences. The broad category includes RAG-based (retrieval-augmented generation) chatbots, AI-powered recommendation widgets, and publisher-branded answer surfaces. The competitive differentiation Taboola claims is its existing scale across the open web, with the company reporting approximately 600 million daily active users across its publisher network, which includes NBC News and Yahoo.
For publishers, the key commercial question is whether on-site AI engagement translates into measurable advertising yield or subscription conversion, not merely longer dwell time. The risk of cannibalising existing pageview-based ad revenue by consolidating reader journeys into fewer, AI-generated response pages is a concern that the release does not directly address.
Regulatory read-across
Generative AI products embedded in news content also carry emerging compliance considerations. The EU AI Act's transparency obligations for AI-generated or AI-assisted content will apply to publishers operating in European markets, requiring clear disclosure when readers interact with AI-generated responses rather than human-authored articles. Wikitree's primary audience is South Korean, but Taboola's wider publisher network spans markets where those obligations will become enforceable from 2026 onwards. Publishers integrating DeeperDive will need to satisfy themselves on disclosure requirements before any European rollout.