Ask FT, the Financial Times’ generative search experience, has now been live for nearly a year, and was officially rolled out to all FT Professional subscribers last month. To learn more about how readers are using Ask FT and how Ask FT is bringing value back to our business, please read my colleagues’ earlier articles. This third and final article in our series explores how we are thinking about developing Ask FT in the future, so as to maximise value to readers and the Financial Times.
The future of Ask FT can be thought of in three parts: 1. Improving it, 2. Evolving it and 3. Learning from it. We spoke with our team that has been driving forward the product development to understand what’s next.
1. Near-term: Updating Ask FT to better meet audience needs
In the near-term, our product teams, informed by audience feedback, are focused on how to enhance the accuracy and user experience of Ask FT. The future of the tool will be shaped by user needs.
Firstly, changes have already been made to make answers more helpful. We made a number of improvements at the end of 2024, resulting in significantly better answers. Most notably, we started rephrasing user queries for better search and performing multiple searches in parallel in order to fetch as much relevant context as possible. We also switched to what was then the most intelligent model available for the final answers (Anthropic's Sonnet 3.5).
Accuracy has been a greater challenge for numerical questions. As Krum Arnaudov, a Senior Data Scientist working on Ask FT, puts it, “At the moment, we tend not to answer highly quantitative questions as this is not what the tool is optimised for… Feedback shows that users actually value us sometimes saying we don’t have the answer and being transparent about that.” That said, we know that readers would like to see more economic and financial data in Ask FT. While technically complex, this could add significant value to users and matches their expectations of questions Ask FT should be able to answer.
Secondly, regarding UX, we are refining the presentation of answers. Readers have asked for formats that are easier to quickly understand and share, such as highlighted keywords, bullet point summaries and sub-headlines. These features will all make it easier to quickly scan for the answer. Other ideas from user interviews include: adding a filter for removing specific types of articles, such as opinion; a stop button to stop answers from streaming and the ability to reorder articles in order of recency.
Questions to ask yourself:
… If your media organisation already has a chatbot: Did the way readers used your GenSearch experience match your expectations? How might the UX be improved to better meet audience needs?
… If you don’t yet: Do you think there is a user need for a chatbot? If so, what do you think your audience would use it for?
2. Long-term: Evolving Ask FT to serve the needs of future audiences
While Ask FT is no longer a novelty, it is still not the primary way our readers interact with the FT.
“Ask FT enables our customers to get trustworthy answers to their business questions. However, it is some ways away from becoming indispensable in how they use the FT… with Ask FT, we must make deliberate product and business strategy decisions to meet users where they are, in terms of how and where they want to use the tool.”
Tanya Brady-HalliganSenior Product Manager, Financial Times
Going forward, a priority will be to further embed Ask FT into the natural user journey, such as the App. In its original form, Ask FT existed on a beta landing page. It has since been incorporated into the homepage but there are other ways we can further embed it to foster a daily habit. Adding Ask FT to the App could give people a way to get context and updates on emerging stories, similar to a summarisation companion.
In the last month, Ask FT was rolled out to the entire Professional subscriber base at the FT: over 1.5 million readers globally. This gives us a lot more data and insight into what readers are using the tool for. We have an opportunity to use the questions our audience asks to inform our content and product strategy going forward. To respect reader confidentiality, we never analyse individual usage data. That said, we can observe trends on an aggregate or industry level. In this way, Ask FT gives us daily insight into what issues are top of mind for our readers. Ask FT gives us an unfiltered look into how the world uses our news.
The evolution of Ask FT is intertwined with the evolution of Generative AI itself. We are living through a time in which technology is evolving at a pace never before. Perhaps, the next frontier of news will be agentic experiences and hyper-personalisation. Our response, at the FT, will be to continue to meet our readers where they are. Ask FT gives us a window into the needs of our most forward-thinking readers: the early adopters. These ‘super users’ continue to give us insights into the changing behaviours of our readers. We are learning from them, knowing that the next generation will be AI-natives and will expect ever higher levels of personalisation and innovation from their news experience.
Questions to ask yourself:
… If your media organisation already has a chatbot: What do you think the next 1-2 years looks like for your product?
To learn more about how your organisation can utilise Ask FT, please visit professional.ft.com/ask-ft.
About the author
