Page Comparison

...

This might be used in order to route inbound queries to language-specific teams or queues - by using a Decision action after this one, you can choose to split your flow into new branches based on the value found in the output of the Detect Language action.

Detect Sentiment

Back to top

What is it?

This action will use AI to detect the sentiment of the interaction (based on the most recent 20 messages), and store the result (i.e. the sentiment detected) in the action output. That action output can be used later in your flow. The possible outputs are: Positive Neutral Negative and Mixed

...

• Custom delays and timers that are beyond the current Start Timer action functionality

• Identifying the dates of the next X working days, and storing the output in variables so you can present them as quick reply options for the customer in a ‘Send Autoresponse’ action

• Converting the result of an HTTP request into an array (list) of variables that you can then use in your flow

And more!

Keyword Filter

Back to top

What is it?

This action is similar to the ‘Decision’ action, but has two critical differences that make it perfect for bulk filters:

• You can add lots of values to each condition - either by manual input, or bulk import.

• If ANY of the conditions for a particular branch is true, the criteria will be met.

More on using Keyword Filters here: Keyword Routing

What is it used for?

Keyword filtering has many different applications - but some examples might include:

• Spam and noise filtering - comment competition running on Facebook? No problem! Filter out all the responses so you don’t waste agent time manually junking.

• Alert filters - look out for high risk words that might indicate fraud or security concerns

• Escalations - scan for highly emotive language (expletives, for example), and route those interactions to a more experienced agent

Make HTTP Request

Back to top

What is it?

This action allows you to make API calls to external systems. There are three primary functions - GET (collect data), POST (send data) and PUT (update data). We’ll store the result of your HTTP request in the Output variable, so you can then reference it in your flow. Any system (CRM, ERP, WFM…the acronyms never end!) can be called, as long as your system is compatible. Get in touch if you want to discuss an integration.

This action is often followed by the ‘JavaScript’ action, to manipulate the raw data you’ve retrieved and sort it into a list of individual variables to use in your flow.

What is it good for?

...

Synchronising your system's customer data with Gnatta so we can streamline your processes

Sending data based on Gnatta events to other systems (BI systems such as Power BI, CRMs, anywhere!)

...

Here’s an example JS action you might use to extract the individual fields from a QA action output, ready to use in your flow (to update data fields or make decisions etc.)

//Parse and log output
const quality = JSON.parse(context.GetObjectJson("Quality.Output"));
context.Log(JSON.stringify(quality));

//Set values as contexts
context.Set("Var.Overall", quality.Overall);
context.Set("Var.Rating", quality.Rating);
context.Set("Var.Escalate", quality.Escalate);
context.Set("Var.Confidence", quality.Confidence);
context.Set("Var.Explanation", quality.Explanation);
context.Set("Var.SolutionFound", quality.Detailed.Solution.Found);
context.Set("Var.EmpathyFound", quality.Detailed.Empathy.Found);
context.Set("Var.IntroductionFound", quality.Detailed.Introduction.Found);
context.Set("Var.Tone", quality.Detailed.Tone);
context.Set("Var.Spelling", quality.Detailed.Spelling);
context.Set("Var.Grammar", quality.Detailed.Grammar);
context.Set("Var.Punctuation", quality.Detailed.Punctuation);

And more!

Keyword Filter

Back to top

What is it?

This action is similar to the ‘Decision’ action, but has two critical differences that make it perfect for bulk filters:

• You can add lots of values to each condition - either by manual input, or bulk import.

• If ANY of the conditions for a particular branch is true, the criteria will be met.

More on using Keyword Filters here: Keyword Routing

What is it used for?

Keyword filtering has many different applications - but some examples might include:

• Spam and noise filtering - comment competition running on Facebook? No problem! Filter out all the responses so you don’t waste agent time manually junking.

• Alert filters - look out for high risk words that might indicate fraud or security concerns

• Escalations - scan for highly emotive language (expletives, for example), and route those interactions to a more experienced agent

Make HTTP Request

Back to top

What is it?

This action allows you to make API calls to external systems. There are three primary functions - GET (collect data), POST (send data) and PUT (update data). We’ll store the result of your HTTP request in the Output variable, so you can then reference it in your flow. Any system (CRM, ERP, WFM…the acronyms never end!) can be called, as long as your system is compatible. Get in touch if you want to discuss an integration.

This action is often followed by the ‘JavaScript’ action, to manipulate the raw data you’ve retrieved and sort it into a list of individual variables to use in your flow.

What is it good for?

• Synchronising your system's customer data with Gnatta so we can streamline your processes

• Sending data based on Gnatta events to other systems (BI systems such as Power BI, CRMs, anywhere!)

  • You can then use this data to analyse or take action however you like

  • We see our clients use our data alongside sales data to help map out forecasts and performance

• Sending user’s ‘states’ to a work force management system to help with schedule adherence

...

This action will use AI to provide a qualitative assessment of the interaction, writing a short but detailed report. Use the ‘Brand Tone’ field to provide specific guidance for the AI (leaving it blank will exclude it form the AIs assessment). You can also choose to toggle additional assessment factors off - though note that the fewer parameters you include, the less accurate the AIs assessment will be. ). You can also choose to toggle additional assessment factors off - though note that the fewer parameters you include, the less accurate the AIs assessment will be.

The report contents will include:

Overall: Very Good
Rating: 95
Escalate: False
Confidence: 90
Explanation: The response was polite, addressed the customer's feedback, and provided useful information on how to get further help. The tone was friendly and supportive.

Detailed Breakdown:
Solution: True
Empathy: True
Introduction: True
Tone: 5
Spelling: 5
Grammar: 5

What is it used for?

This action might be followed up by an Add Note action to append the output of the Quality Assessment to the interaction - or a Decision action JavaScript action to split out specific parts of the report into variables ready to take further steps for high risk interactions based on keywords detected in the Quality Assessment!. (Such as updating custom data fields, or splitting flows depending on whether Escalate = False etc.)

Release Thread Control

Back to top

...

Similar to the AI Assistant available to agents for the interactions radar, this action will use AI to compose a possible response to the interaction, based on the most recent 20 messages and any available text data fields. The suggestion will be stored in the action output, where it can be used later in your flow. later in your flow.

The output of this action will be formatted as follows:

  "Timestamp": "2024-11-11T10:00:00Z",
  "IsResponse": false,
  "Message": "This is a sample message."

...

Similar to the AI Assistant available to agents for the interactions radar, this action will use AI to summarise the interaction, based on the most recent 20 messages and any available text data fields. The summary will be stored in the action output, where it can be used later in your flow.

The output of this action will be formatted as follows:

  "Timestamp": "2024-11-11T10:00:00Z",
  "IsResponse": false,
  "Message": "This is a sample message."

...