Sentiment Analysis
One of three words for the whole conversation, decided in the same reading as everything else. It is a stamp on a finished call, not a line rising and falling while somebody is still talking.
Sentiment on the day's calls
stamping- Dana WhitlowSupport4:12negative
- Ben OseiNew business2:38positive
- Priya RamanBilling3:04neutral
- Tom FerrarSupport7:05negative
- Ray WhitfieldBilling0:11analysing
- Alina CruzNew business3:19analysing
- Marta LangSupport5:52analysing
Product figures from the platform’s own defaults - not customer averages
How it works.
Three words, matched rather than trusted
What comes back is trimmed, lowercased and then matched against positive, neutral and negative. A fourth word is not filed under protest: it fails, the reading repeats, and the column keeps whatever it already held.
Whole call, single value, its own column
Nothing is produced per speaker, per minute or per turn. There is one field, twenty characters wide, holding one word, which is exactly why it can be filtered on directly instead of being worked out again at query time.
It becomes a control on the call list
The list takes one of the three alongside the analysis state and a score floor, resolves the analyses that match, then narrows the calls to those. That lookup is deliberately bounded, so a very wide range answers with a bounded set.
Counted for each person overnight
An overnight rollup tallies the three values per agent for the previous day and also averages them numerically, positive at one and negative at nought. That is what lets a fortnight read as one figure per person without walking every call again.
One moment in every read.
Every read passes through the same seven. Sentiment Analysis 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- Values
- positive, neutral, negative. Case and whitespace forgiven, a fourth word is not
- Granularity
- One reading for the call. Nothing per speaker, per turn, or across its length
- Very short calls
- Take neutral by instruction rather than a guess at what was probably meant
- Filtering
- On the call list, combined with the analysis state and a minimum overall score
- Lookup ceiling
- Five thousand analyses in one pass on the sentiment and score filters, fifty thousand on the not-yet-analysed one. Past either, the answer is quietly short rather than flagged
- Not the same as
- How the interaction landed. Tonality is how each side sounded saying it
- Alerting
- None. A negative is something you ask for, not something pushed at anyone
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.
One reading of the transcript returns the summary, the sentiment, the tone, what each side intended and a set of scores, in a shape fixed in code.
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.
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.
Four of six speech paths return the channel each side was recorded on, so the transcript knows who spoke rather than guessing, and one control swaps a pair.
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 sentiment analysis on your own floor.
Thirty minutes, your numbers and your data. We will set sentiment analysis up live and you can decide from the thing itself rather than from this page.
14-day trial · no card · migration included