BURKI FINAL TRANSCRIPT TXT EXPORT REVIEW
Prepared October 17, 2026. Free worksheet; production usage is separate.

Purpose: review permitted saved final text for one specific question.
This is a planning resource, not a private transcript or proof of a call test.

1. AUTHORIZE THE REVIEW
Organization/workspace:
Reviewer and approved purpose:
Call ID (keep private where required):
Question the transcript should answer:
Approved local storage/sharing location:
Applicable retention/access policy:

Do not export a record you are not authorized to handle. The TXT file does
not automatically redact personal content. Do not alter retention or policy
just to make a download available.

2. USE THE CURRENT LIST CONTROL
Open Calls -> Call history.
Choose Search type: Call ID, enter Search call ID and select Apply.
Check the intended row using information you are authorized to see.
Select its Download transcript icon. The current list requests TXT.
The implementation's local filename is call-{id}-transcript.txt.
Check the actual browser result; do not treat a success message as completeness.

The list control was observed October 17. No private record was opened or
exported for this resource. Content/format behavior is source-verified.
Do not infer a CSV/JSON dashboard chooser from other API capabilities.

3. INSPECT THE PERMITTED ORIGINAL
Export date/time and known timezone:
Actual filename:
Header call ID matches intended record: yes / no / unknown
Header start value checked against authorized context:
Conversation lines after header: yes / no
Expected final turns or coverage, and why expected:
Observed final turns or coverage:
Missing/surprising content:
Unknown or questionable speaker labels:
Unchanged original stored under approved process:
Annotations kept separately:

TXT contains saved final rows, ordered by stored creation time. Interim
recognition text is excluded. Header-only output is possible if there are
no final rows. Missing text does not establish that no one spoke.
The row's stored speaker is used; a missing label becomes Unknown.

Per-line HH:MM:SS comes from row creation time. It is not guaranteed audio
alignment and lacks a per-line date/timezone. Do not calculate caller-heard
latency from it or assume it matches a dashboard's local representation.

4. SYNTHETIC EXAMPLE, NOT AN ACCOUNT EXPORT
The following invented lines illustrate a correction:
[10:14:02] user: Please use the north entrance.
[10:14:07] user: Sorry, I mean the south entrance.
[10:14:10] assistant: You mean the south entrance.

Narrow finding: in this illustration, retained text includes the correction.
Unknown from text alone: whether the caller heard the response; whether any
business system or delivery instruction was updated.
If only the first line were present, investigate expected retained coverage.
Do not conclude that a correction was never spoken.

5. KEEP EVIDENCE STATES SEPARATE
Observed saved text:
Reviewer interpretation:
Facts verified outside the TXT, with authorized evidence reference:
Audio/playback evidence, if the question concerns audible output:
Action/authoritative business record, if the question concerns completed work:
Unverified explanations or missing evidence:
Next responsible reviewer and approved channel:

A transcript statement about success is not an action receipt. TXT is not
a recording, a full call backup, or evidence of caller identity. Preserve
uncertainty instead of assigning an Unknown turn to a person by guesswork.

6. FAILURE REVIEW
Failure/error exactly as observed:
Account/organization/call ID checked:
Permitted retention/final-row availability checked by authorized operator:
Support reference (avoid attaching private content by default):
Outcome: finding supported / more evidence needed / access restricted

No actual private download, voice call or provider action was performed to
prepare this worksheet. Use the current approved access and review process.