Change Log

Bug Fixes

• Interviews run to the end

  • A phrase such as "we're done with this section" could be mistaken for the end of the interview, so the last question was never asked and no closing was spoken
  • A participant who stays quiet is now asked the question again, instead of the interview moving past it
• Quotes are the participant's own words

  • The Quotes tab in Insight evidence could fall back to the analysis's own third-person summary when it had no quote to show; it now shows only what the participant actually said, and leaves out anything it cannot quote verbatim
  • Answers produced before this change are re-checked when you open them, so an older answer cannot surface a paraphrase as a quote
• Analysis no longer looks empty when responses exist

  • Transcripts saved in older formats were skipped when the overview was built, so a study could show participants and a completion rate beside empty analysis panels
  • Where the tiles and the panels genuinely disagree — responses recorded but no transcript text to analyse — the dashboard now says so instead of leaving a blank
• A completed interview means the questions were answered

  • Voice sessions were marked complete whenever they ended, even when no questions had been answered; completion now follows how the interview ended and how much of the guide it covered
• Participant entry

  • A closed study says so on the first screen, instead of after the participant has given their details, chosen a mode, granted camera and microphone and sat through the countdown
  • Preview mode is labelled from the first screen, so the welcome, screener and permission steps no longer look like a live session
  • Opening a response that has already finished explains what happened, rather than showing a generic error with a Try again button that could never work
• Study setup

  • The steps in the setup sidebar are clickable again — when the sidebar was expanded they rendered as plain text and did nothing
  • Multiple-choice questions written by Convo AI now come with their options; they previously arrived blank, which failed validation and blocked saving the guide
• A study that is out for review now shows "In Review", rather than every study reading "Collecting Responses" from the moment it was created
• Ranked questions are read out as a numbered list in voice interviews, matching how they appear in chat
• Recently Viewed reflects the studies you actually opened, instead of repeating the first three in the list below it
• Sign-up refuses trivially weak passwords such as a single repeated character or a straight run of digits