Final transcription after a call

When a real voice call closes, BeAI queues a final transcription in the background. The selected model receives only the caller audio track and the existing call summary as context. Live text remains available while processing is pending.

After success, caller text is replaced in the conversation, API detail and text export. Bot replies, their dates and their original order are preserved. The call summary and completed actions are not rewritten.

Processing mode

Platform administrators open Platform settings → Final call transcription and choose Standard mode or Alternative mode. Technical service names remain hidden. Changes apply to newly closed calls; queued jobs keep their saved mode. Organisation administrators cannot change this platform setting.

The equivalent operation is PATCH /api/v1/system-settings/final-transcription; GET /api/v1/system-settings returns data.final_transcription. Its technical schema and allowed values are restricted to platform administrators.

Retained and temporary recordings

All real voice calls are captured for this processing. When bot recording is enabled at call start, the original recording is retained and remains available to authorised users as before.

Otherwise, audio stays in private temporary storage, with no player, download or audio API access. It is deleted after successful transcription. When the bot's auto-improvement also requires the recording, transcription waits for that audio analysis to finish. Temporary files are retained for at most 24 hours if processing fails persistently.

The bot recording option therefore controls retention and playback. Simulations, written WhatsApp messages and voice notes do not use this queue. Historical calls are not automatically reprocessed.

Processing states

The conversation detail shows Queued, Processing, Retry scheduled, Completed, or Failed. Reload the page to refresh it. The API exposes the same state in the conversation's final_transcription field. A failed job keeps the live transcript and displays a diagnostic code without sensitive content.

Temporary failures have up to five attempts with increasing delays. An interrupted worker resumes its durable queue. Saved results for completed audio chunks are reused. Text exports always contain the currently visible transcript.

Reading limits

Caller text is placed between bot replies by matching it to historical live turns. Placement remains approximate when the live transcript was inaccurate. Returned words are not manually corrected. Historical times are message log times rather than precise speech boundaries. Timestamp precision depends on the processing mode; some modes provide no word timestamps.

The existing summary can contain errors and influence name recognition. Neither model guarantees a completely accurate transcript.

Purchase estimates

Purchase estimates are hidden by default. Platform administrators may temporarily enable Purchase estimates from the avatar. The conversation breakdown then shows processing, quantities, unit prices, currencies and subtotals. The preference is stored in neither a cookie nor the database and resets when the browser closes. Closing all BeAI tabs or changing pages may also reset it depending on browser support.

Estimates include final transcription and optional auto-improvement without changing customer charges. Depending on the mode, quantities are audio seconds or tokens; duration is never converted into invented token counts. Reused results are not counted twice. Unrecorded responses, missing rates or incompatible currencies can make the estimate partial. Each response keeps its frozen historical rate.

GET /api/v1/conversations/{id}?include_purchase_estimates=true exposes estimates to platform administrators only. Without opt-in they remain hidden. The overall estimate already includes final transcription; do not add the subtotal twice. Purchase-rate configuration remains restricted to the platform.