Ask FT: Your direct route to insight

How Ask FT is fulfilling user needs: One year in with our generative AI feature

Last March, the Financial Times launched its own proprietary generative AI feature, Ask FT. The search tool was developed with the objective of creating new experiences for readers by using archive content to generate answers to user questions.

One year into launch, we went behind the scenes to understand how the tool has evolved. We connected with the development team to learn a bit more about user adoption, how Ask FT is being integrated in day-to-day workflows, and what’s next on the product roadmap.

Who uses Ask FT?

Ask FT was first launched internally within the FT to gather feedback, then beta tested with a subset of FT Professional customers. As of April 2025, Ask FT became accessible to all FT Professional customers. We’ve received positive feedback from users who trust the product’s outputs and find immense value in leveraging the tool to go deep on industry issues, trends, events or stories.  

Based on usage data, the majority of users work in sectors such as finance, consulting and legal. These customers have used Ask FT for topic exploration, looking to understand updates and developments on events as they arise. This aligns with what the team was aiming to achieve: building a tool to answer user questions about topics that the FT covers. Commonly cited use cases included assisting with preparation for client meetings or reports, providing expert opinions or perspectives, identifying reliable data sources, and staying up-to-date with developments.

Putting Ask FT to work

Initial research on user behaviour towards generative search engines has shown that users have a tendency to take generated responses at face value, without cross-checking references. Ask FT users, however, are the opposite. Our team has seen that users tend to dig into sources and refine prompts to return exactly what they need. Our users want to read, react, and easily get the full story behind a given stat or number. 

During the beta test, the team also uncovered a variety of alternative use cases that users were integrating into their day-to-day workflows. Tasks such as fact-finding, article summarisation, and specific article search came up as top use cases for users who implemented Ask FT into their ways of working. Our pilot users utilised the tool for information retrieval, seeking answers to questions on topics such as the US election, major geopolitical events, performance of financial markets or specific companies, and green energy and sustainability. 

The case for trust over speed

Prior to development, the team agreed to prioritise credibility and output quality over speed. A competitive analysis of other generative search engines in the market revealed that when built for speed, output responses often failed to return credible data. Given brand trust is a cornerstone of the FT’s journalism (and is our key differentiator as a publisher) it was paramount that user trust and development transparency were at the forefront of the team’s product development strategy.

The tool works by serving a well-defined niche in the generative search landscape rather than being a ‘catch-all’ tool. Whereas search tools like ChatGPT, Google AI Overviews, and Perplexity function as search and answer engines, Ask FT acts as an assistant for sifting through aggregated content. The engine is purpose-built to be trained exclusively on FT content, without incorporating external sources or prior tool knowledge. 

Ask FT offers a trusted perspective, particularly important within an industry-wide environment which can exhibit low levels of trust towards mainstream media and AI technologies. The tool is exclusively powered by trusted reporting from the FT newsroom. Users know they are getting answers driven by fact-based journalism and cited sources.

Our product teams looked to maintain transparency by providing disclaimers in user response boxes and taking responsibility for outputs that may fall short of editorial standards. Part of this responsibility is being explicit in the limitations of the tool, ensuring Ask FT calls out a lack of sufficient information to return a credible answer. 

Building products around user needs

Ask FT fundamentally changed the way our product teams build with AI. Dilyana Evtimova, a Senior Product Manager at the FT, wrote in a blog post that “... while the AI technology is new, user needs rarely change overnight”. As such, the team has reframed the product development process in two key ways: focusing on user needs and making space for constant iteration. 

The best product strategies start with problems and opportunities in mind, then use tools to execute or scale solutions. Ask FT started with identifying a clear user problem: FT Professional subscribers were looking for specific sources or data points, and struggled to find them quickly. The solution was then mobilised by AI rather than treating AI as the end goal. By focusing on the need before the technology, we’re able to drive meaningful business value and implement practical solutions for readers. 

Within FT Strategies, we’ve seen strong results helping publishers build chatbots grounded in clear user needs and use cases. Many have developed semantic search tools to achieve key business metrics, including driving registrations or increasing overall time on site. Our focus on KPI-driven development has led to meaningful, measurable results across our engagements. 

Second, constant iteration became incredibly important to product development. Given the plethora of data inputs and features AI models need to run, requiring additional assumptions than a typical product feature. Consequently, receiving feedback and re-training tools to incorporate missed use cases becomes an essential component of the development process to help maximise value. 

To learn more about how your organisation can utilise Ask FT, please visit ft.professional.com.

If you're looking for support in shaping product and development strategy, or want to understand if building a content-powered conversational AI product is right for your business, please get in touch with our expert team.