Event Library
Browse a prototype catalogue of FT Live-style events. Facets make it quick to narrow the collection by year, country, theme and speaker.
Loading transcripts...
Loading events...
FT Live Conference Workbench turns conference transcripts and audio into a searchable research surface for editors, producers, analysts and event teams.
Browse a prototype catalogue of FT Live-style events. Facets make it quick to narrow the collection by year, country, theme and speaker.
Search across all conference sessions from the landing page. Results show matching transcript passages and jump straight into the relevant session.
Read each session as timestamped transcript chunks, with the selected event context preserved while moving between sessions.
Play the session audio in the workbench. Timestamp links, search results, chapter markers and individual transcript words can seek to the matching point in the audio.
As audio plays, the transcript highlights the current neighbourhood of words so it is easier to follow along without automatic scrolling.
Each session has approximate chapter markers with timestamps. These provide quick entry points for scanning long conversations.
Transcript chunks include approximate speaker labels. Clear introductions are used where available; otherwise the prototype falls back to safe labels such as Moderator, Panellist or Guest.
Likely people, organisations, places and other named entities are highlighted inline to make dense transcripts easier to scan.
Generate session-level editorial analysis with OpenAI using configurable prompt templates such as top themes, interesting moments and editorial highlights.
Timecodes in transcript views and generated LLM output are converted into links so evidence can be checked against the audio quickly.
LLM responses can be cached locally or in the Cloudflare deployment cache layer, reducing repeat calls and speeding up repeated analysis.
The app is designed for Cloudflare Pages, with server-side functions for OpenAI calls, cache clearing and GitHub-backed audio proxying.