Guide · Capture

How the call gets in decides what is possible.

Every conversation about automating call work eventually becomes a conversation about capture. Get this decision right early, because it sets the ceiling on everything downstream.

All guides

The three tiers, side by side

The rest of this page is the detail. This is the decision.

The three capture tiers compared on what reaches the system, what becomes possible, what it costs to set up, and who has to be in the room.
  Stored recording Live stream (SIPREC) Locked box
When the audio arrives After the call endsWhile the call is happening Either, inside your boundary
During the call NothingPrompts, lookups, corrections Depends on the tier you run inside it
After the call Full draftingMostly written alreadyFull drafting
Setup cost Very little. The recordings already exist Real telephony work on your SBC The most of the three, and the slowest to change afterwards
Who has to be involved Whoever owns the recording store Someone who knows your telephony Your security function, first
Choose it when You are starting. Most floors should start here Something has to happen before the customer hangs up The data is not allowed to leave. Not because it sounds safest

Consent law sits underneath all three and is the section below.

Tier one: the stored recording

You already keep recordings. Something reads them after the fact and produces the record. This is the simplest thing that works and it is where most floors should start.

What you give up. Nothing can happen during the call, because the call is over. No live prompt, no mid-call lookup, no correcting a wrong field while the customer is still on the line.

What it costs. Very little. The recordings exist, the retention policy exists, and the connection is a read.

Tier two: the live stream

SIPREC is the IETF standard, defined in RFC 7866, for streaming live call media to a recording server. If your telephony supports it, the audio can reach a system while the call is happening rather than after.

What you gain. Anything that has to happen before the customer hangs up. The record can be most of the way written by the time the call ends.

What it costs. Real telephony work. Someone who knows your SBC has to be involved, and the answer to "does our platform do SIPREC" is often "yes, on a licence we do not have".

Tier three: the locked box

Where the data cannot leave a boundary, the processing comes to the data instead. The system runs inside your environment and nothing crosses the line.

What you gain. The conversation with your security function gets much shorter, and in some lines this is the only structure that is allowed at all.

What it costs. The most setup of the three, and the slowest change cycle afterwards. Choose it because you must, not because it sounds safest.

The consent question underneath all three

Recording-consent law in the United States varies by state between one-party and all-party rules, and roughly a dozen states require all parties to consent. If you take calls across state lines, and almost every floor does, your policy has to be built for the strictest state you touch rather than the one you are sitting in.

This is general education and not legal advice. The specific question of what your disclosure has to say, and when it has to be said, belongs with your counsel before any capture decision is final.

Related: what "provenance, not generation" means

Want this looked at on your own floor?

Twenty minutes on your systems and your numbers. You leave with the map either way.