Interview replay shows corrupted AI‑test session and playback is broken

Bug: Interview replay shows corrupted AI‑test session and playback is broken

Context
Study: Workforce Engagement & Retention Study
Interview: “Max”
Env: ontora-git-max-refactor-ontora-icon-agent-3gc.staging.ontora.com

Issues

1. Wrong / corrupted content
   The replay is clearly an AI‑UI test, not a real interview:

   • Participant talks only about the agent icon and voice animation, e.g. “when the agent is talking, the icon should move…” and “make the agent animation voice look kind of more natural…”

   • The agent keeps re‑starting intro questions (“walk me through a typical day…”) and sending multiple closing messages.

   • Summary already says this was “testing or evaluating an AI agent interface” rather than an engagement interview.

2. Broken playback UX (jumping + not scrubbable)

   • The player shows 0:27 / 0:27 but:

     • The timeline is not scrubbable: dragging or clicking on the timeline does not seek reliably (or at all).

     • When you try to move within the replay, it jumps unexpectedly or snaps back, instead of smoothly seeking to the selected position.

   • This makes it impossible to review specific moments in the conversation.

Expected behavior

• This test‑like session should either:

  • Be excluded/flagged as a test, or

  • Not appear as a valid interview in the Workforce Engagement study.

• The replay player should:

  • Allow smooth, accurate scrubbing/seek along the timeline.

  • Not jump or snap unexpectedly when the user drags or clicks the seek bar.

  • Keep transcript and audio in sync when seeking.

Actual behavior

• A short AI‑UI test session is surfaced as a full “Max” interview in the study.

• The replay timeline behaves unreliably:

  • Scrubbing is effectively disabled or very buggy.

  • Seeking causes jumps / resets instead of precise navigation.

• Together, this makes the interview unusable for analysis.

Steps to reproduce

1. Open the Workforce Engagement & Retention Study in staging.

2. Open the “Max” interview detail page and start the replay.

3. Try to:

   • Scrub or click at different positions on the 0:27 timeline.

   • Jump back and forth to specific points.

4. Observe:

   • The replay either doesn’t seek at all or jumps around / snaps back.

   • Transcript + summary clearly describe a UI‑test conversation, not a normal interview.

Impact

• Pollutes study data with a non‑interview session.

• Makes it impossible to properly review the audio/transcript due to broken scrubbing.

• Undermines trust in both replay and study report quality.

Acceptance criteria

• This “Max” test session is either:

  • Excluded from the study and clearly labeled as a test, or

  • Correctly replaced with the intended interview (if one exists).

• Replay component:

  • Scrubbing works reliably for this and other sessions.

  • No unexpected jumps or resets when seeking.

  • Transcript and audio remain in sync after seeks.

You can tweak naming/formatting to match your existing Featurebase posts, but this covers both the corrupted content and the broken scrubbing/jumping behavior.