On Saturday, @TechCrunch included something remarkable inside one of its stories: the world's first interactive avatar of a reporter built by @synthesiaIO and capable of answering questions about a story that the reporter has written: techcrunch.com/2026/09/26/i-c…
Readers can talk to an interactive version of @DominicMadori directly from the article, asking deeper questions about her reporting on VC-funded startups engaging in fraud. When the avatar pulls information from its given resources (two academic papers and Dominic's article on the topic), it can slightly rephrase them for clarity, but it cannot go beyond what the original material says. And if it cannot find an answer in its knowledge base, the interactive avatar says it does not know rather than trying to guess or make up an answer.
In my view, this opens up new ways for readers to interact with a story and the reporter who wrote it. Most online journalism today is still a one-way transaction: a reporter spends weeks gathering information, compresses it into an article, and the reader gets whatever made the final cut.
An interactive avatar creates another layer on top of that reporting. Readers can ask more questions about the article, follow their own line of curiosity and get answers grounded in the work the journalist has already done, including background information and fun facts that simply could not fit into the written story.
For me, this TechCrunch experiment is an early example of what that can look like when the avatar is correctly constrained, embedded in context and built around a specific piece of work.
00:00
