“The export button is broken - when I click it, nothing happens and there’s no error.”
Reproduces on Chrome 121 + Safari 17. Same step in dashboard. Audio confirms.
Fireflies competes on breadth of meeting assistance. Citesvue competes on whether a single claim can be defended, and on whether the screen and the documents are in the same answer.
“The export button is broken - when I click it, nothing happens and there’s no error.”
Reproduces on Chrome 121 + Safari 17. Same step in dashboard. Audio confirms.
Their description is quoted from their own pages. Ours describes what the deployed product does today.
Fireflies.ai calls itself "The #1 AI Assistant For Your Meetings", publishing 95% transcription accuracy across 100+ languages with speaker recognition, an assistant called AskFred for questions about meetings, conversation intelligence such as talk-time, sentiment, and topic trackers, and "200+ AI Skills" layered on top of notetaking.
Citesvue narrows the surface deliberately. One index holds transcript, keyframes, and documents; one discipline governs answers, which is that every claim binds to at least one citation you can click. Breadth arrives as more kinds of evidence, not more kinds of summary.
The rows that decide this are not about summary quality. They are about what a colleague can verify without reopening the source.
Read from Fireflies.ai's own public pages on 4 August 2026. Rows marked Not published mean the capability is not described there, not that it is impossible. Products change; if we have a row wrong, tell us and we will correct it.
A comparison that never concedes anything is marketing. These are the situations where we would tell you to stay.
The fastest honest test is a recording where something went wrong on screen. Ask both tools what the error said. One answers from what was spoken; the other can answer from what was displayed.
Any recording you already have. No bot, no calendar connection, no import job.
The specification, contract, or test plan the session was about. PDF, Word, or Markdown.
The one where the answer matters. Check whether what comes back can be verified without reopening the source.
Looking for an Otter.ai alternative? Compare transcription and meeting summaries with citation-bound evidence from speech, screen, and documents.
A Fathom alternative for evidence work: frame and page citations, documents alongside recordings, and artifacts that arrive in Jira with their proof.
A tl;dv alternative for evidence-grade output: cited answers across recordings and documents, frame-level proof, and exportable evidence packs.
Or it could sit in a Drive folder nobody opens again. The difference is whether it has citations attached.