Call recording & transcripts

Every call an assistant handles comes back to you in writing — transcript, summary, and outcome — and if you switch recording on, the audio comes with it.

What it is#

Every conversation a OnCore assistant handles produces a written record automatically: a full transcript, a turn-by-turn message log, an AI-written summary, and structured outcome data. Audio recording is a separate switch you set for each assistant — when it's on, every finished call also carries a link to its audio recording. Turning recording off never affects the written record: transcripts, summaries, and analysis are produced for every call regardless.

What you can do with it#

• Quality assurance and coaching — review exactly what was said on any call, in text or audio, without sitting on live calls.
• Dispute resolution — pull up the recording and transcript of a specific booking, order, or complaint when a customer says "that's not what I was told."
• Sync outcomes into your systems — receive the transcript, summary, and recording link in your CRM, ticketing, or data warehouse the moment a call ends, automatically.
• Reporting and analysis — every call is summarised and categorised with labels you define (for example booking, edit, cancel), so you can measure what your callers actually ask about.
• Compliance review — keep an auditable record of what your AI assistant said to customers, per assistant and per call.

How to set it up#

Everything is configured through OnCore's REST API. Recording is a single on/off setting on each assistant — choose it when creating the assistant and change it any time with an assistant update; nothing changes for callers either way, and transcription happens on every call with no setup at all. To retrieve results, pull any call from the calls API — the record includes the transcript, message log, summary, analysis, and recording link — or have the end-of-call report webhook push the same data to your server the moment each call finishes.

Limits & FAQ#

• Recording is per assistant, not per call. The setting applies to every call that assistant handles; there is no caller-facing opt-in or opt-out during a call.
• Bulk export returns lean rows. The bulk export endpoint omits transcripts, messages, and recording links — fetch individual calls for the full payload.
• Webhook delivery is not retried. If your endpoint is down when the end-of-call report is sent, poll the calls API to backfill — the call record itself is always there.
• No recording, no link. Calls that ran with recording off simply carry no recording link; the transcript is still complete.

Do we get transcripts if we don't enable recording? Yes. Transcription, summaries, and call analysis happen on every call regardless of the recording setting.

How soon after a call are the recording and transcript available? When the call finishes — the end-of-call report fires at that moment, and the same call record is readable from the API.

Can we record only some of our lines? Yes. Recording is set per assistant, so if different phone numbers are answered by different assistants, each can have its own recording policy.

What about recording-consent laws? Recording rules vary by jurisdiction, and complying with them (for example, notifying callers) remains your responsibility. Because the setting is per assistant, you can disable recording wherever your policy requires it — transcripts remain available either way.

Can recordings and transcripts be deleted? Yes — call data can be deleted on your own retention schedule, and each cleanup run is reported through Retention audits so deletions stay auditable.

Do chat and SMS conversations get the same treatment? Messaging channels have no audio, but every conversation keeps a full message log and works with the same read APIs — see Conversations.