The analyses endpoint uses our Deep Insights API behind the scenes to determine the appropriate information, advice and feedback to send for a conversation. However, the analysis is performed only on the overall results of the analysis and not each of the intervals. Performing analysis on the intervals is possible, but requires a more intentional approach -- which is why the separate /insights API is made available.
The Deep Insights API takes the result of the conversational analysis -- rather than a conversation itself -- and returns the output. This makes it possible to separately retrieve the API results for a specific interval or even compare two (or more!) people or intervals.
Let's try the second example: comparing two people.
Comparing Two Individuals
The "collaboration" context is a powerful module that compares two people and provides advice for how the two can work together more effectively. Let's say we received the following analysis result for a phone call, between a sales person and their customer, respectively:
{
"values": {
"ranked": [
"data",
"order",
"instinct",
"community"
],
"confidence": 1
},
"sensory": {
"ranked": [
"visual",
"auditory",
"kinesthetic"
],
"confidence": 0.88
},
"decision_making": {
"intent": {
"label": "undecided",
"score": 0.281
},
"buckets": {
"low": [
"logic",
"commitment",
"motivation"
],
"high": [
"need",
"ability"
]
},
"confidence": 0.71
}
}
{
"values": {
"ranked": [
"instinct",
"data",
"order",
"community"
],
"confidence": 1
},
"sensory": {
"ranked": [
"visual",
"kinesthetic",
"auditory"
],
"confidence": 1
},
"decision_making": {
"intent": {
"label": "completely committed",
"score": 0.969
},
"buckets": {
"low": [],
"high": [
"need",
"logic",
"ability",
"commitment",
"motivation"
]
},
"confidence": 1
}
}
The Deep Insights API allows me to compare these two individuals -- these two analysis results -- and retrieve some advice about them. And by passing the "collaboration" context, I would expect to get back the "working_with" insights.
I'll pass each of these results to the API in the compare parameter:
POST /conv/v1/insights
X-API-Key: YOUR_API_KEY_HERE
Content-Type: application/json
{
"compare": [
{
"taxonomies": {
"values": {
"ranked": [
"data",
"order",
"instinct",
"community"
],
"confidence": 1
},
"sensory": {
"ranked": [
"visual",
"auditory",
"kinesthetic"
],
"confidence": 0.88
},
"decision_making": {
"intent": {
"label": "undecided",
"score": 0.281
},
"buckets": {
"low": [
"logic",
"commitment",
"motivation"
],
"high": [
"need",
"ability"
]
},
"confidence": 0.71
}
},
"nlg": {
"pov": 2,
"slots": {
"info": {
"first_name": "Randy"
}
}
}
},
{
"taxonomies": {
"values": {
"ranked": [
"instinct",
"data",
"order",
"community"
],
"confidence": 1
},
"sensory": {
"ranked": [
"visual",
"kinesthetic",
"auditory"
],
"confidence": 1
},
"decision_making": {
"intent": {
"label": "completely committed",
"score": 0.969
},
"buckets": {
"low": [],
"high": [
"need",
"logic",
"ability",
"commitment",
"motivation"
]
},
"confidence": 1
}
},
"nlg": {
"pov": 3,
"slots": {
"info": {
"first_name": "Sierra"
}
}
}
}
],
"contexts": [
"collaboration"
]
}
The nlg parameter next to each of the analysis results is the NLG Options object, and includes pov (point of view) and any additional slots that you want to provide. The slot values you provide here will be used to help fill in key slots in the NLG content; you can work with the Cyrano.ai team to determine which slots are recommended for the contexts you're retrieving.
Deep Insights Results
{
"working_with": [
"People who use this communication style are natural collaborators who make very good team players.",
"Randy has an excellent feel for Zoom. They feel comfortable in front of the camera. It is good to keep them engaged, they can zone out if the same person talks for too long. They prefer having a lot of different visual stimulation to hold their attention.",
"Randy prioritizes the way things look and physical action more than auditory distractions. Allow them to make things look good as they craft an action plan.",
"Randy is more sensitive to tone of voice than most other people."
]
}
Wrapping up...
The collaboration module is just one example. Other use cases may call for comparing different analysis results, such as two (or more!) intervals, behavior of a person in two different meetings or different contexts, and more. The Cyrano.ai linguistic team can help your team create new modules for your implementation as well.