Call Summaries
One to six sentences on the record saying why the call happened, what was covered and what happens next, so whoever is reading the list does not have to open the audio to find out.
Which calls get a summary
written as the recordings landDana Whitlow called about an invoice issued at the previous rate. The agent agreed to reissue it today and to raise a credit for the difference rather than carry it forward. A renewal call was requested for Thursday with their finance director joining, and the current terms will be sent beforehand.
Summary written onto the call, one to six sentences of plain text.
Product figures from the platform’s own defaults - not customer averages
How it works.
It is a job behind the call, not a wait in front of it
The recording finalises, work is queued against it, and the rep has already moved on. The identifier is the media type and its row together, so a telephony hook that fires twice produces one run and one charge.
Five gates settle it before anything is spent
Platform switch, workspace switch, whether this kind of media is included, whether the workspace is on automatic, and the length floor. Each is a plain function of settings, and the whole ladder is read once, as the work is queued, so nothing is spent on a recording that was never going to be read.
An unknown length is refused rather than assumed
Where a floor is set and the duration has not been written yet, the automatic path denies instead of reading something that might be under it. The earlier build let that case through, and denying is both the correct behaviour and the less flattering one.
A thin recording gets a neutral answer, not a fuller one
The instruction says outright that a short or oddly phrased transcript is still analysed and that anything it cannot reasonably infer takes a neutral default. What comes back is prose, held on the record as text.
One moment in every read.
Every read passes through the same seven. Call Summaries is the lit one, and everything either side of it is a different page in this category.
- 01Source
the call or the thread it reads
- 02Transcribe
audio into words, with speakers
- 03Read
the pass over the whole of it
- 04Judge
the score, the sentiment, the intent
- 05Extract
the fields and follow-ups pulled out
- 06Write
what lands back on the record
- 07Review
a person checking the machine
The specifics.
7 facts- Length
- One to six sentences of plain text. No code fences and no preamble around it
- Covers
- Why the call happened, the points raised, the outcome, and any next step
- Runs on
- Calls and voicemails, and separately on mail threads. Nothing else has a source type
- Length floor
- Set on the preset. An unmeasured duration is denied on the automatic path
- When speech fails
- Everything downstream is skipped and the record shows the failure and its reason
- Not the same as
- Prose for a person. Intent detection answers the same call as labelled lists
- Editing one
- Not offered. Running the analysis again replaces it, and the change entry records that it happened while carrying neither version of the text
What it reads, and what reads it.
Nothing here invents its input. These are where the material comes from, and where the verdict goes afterwards.
More in Intelligence
12 capabilitiesAI that proposes edits to the record - an insight becomes a field once you accept it.
The model proposes record updates with its confidence and the quote it heard them in, shown as a before-and-after you accept or reject row by row.
Chosen words rather than a fixed list for how the agent sounded, how the customer did, and the conversation overall - the manner behind the sentiment.
Communication, resolution, satisfaction and compliance scored nought to ten beside a separately judged overall, on every call that clears the analysis gates.
Compliance scored nought to ten against what the vertical says matters, because no phrase list exists to recite. A missing reading is excluded, never passed.
What the customer was actually asking for and what the agent committed to, extracted as structured intents rather than left inside the transcript.
Callbacks and meetings pulled out with their times, a confidence figure and the words they came from, ready to accept in one click or reject in one.
Collections, healthcare, insurance, admissions and eleven more - each shaping what the analysis looks for without letting anyone break the output shape.
Threads summarised and read for intent and urgency, but only when a participant matches a record, so the model is never called on mail that is not yours.
Draft from an instruction or rework a message you have, landing in the composer with a suggested subject and going nowhere at all until you send it.
Describe what you track or paste a spreadsheet's columns, and get a named field group back with types, options and placeholders to approve one at a time.
A sentence becomes a grid filter and a sort order with the reading explained back, on the toolbar of every grid that carries a column menu of its own.
OpenAI, Azure, Anthropic, Gemini, Bedrock and ten more, bound per purpose with an ordered fallback chain, a circuit breaker and a cost line on every attempt.
The rest of the platform.
Five more categories, all on the same record and the same bill. Each card names three of its capabilities, so you can tell from here whether it is worth opening.
Intelligence
See call summaries on your own floor.
Thirty minutes, your numbers and your data. We will set call summaries up live and you can decide from the thing itself rather than from this page.
14-day trial · no card · migration included