Cognigy.AI provides creators with customization settings that allow the agent processes to be adjusted to achieve optimal performance.
The Agent Settings are used to configure:
- The Flow Logic of all Flows within this agent
- The Timezone of the agent
- The Intent Mapper of Flows NLU
- The Thresholds for NLU recognition
- The Analytics of the agent
Within your agent, from the left toolbar click Manage → Settings to access the agent settings menu.
General Flow Logic || Flow Settings¶
Additional confirmation words¶
Additional negative confirmation words¶
Continue main Flow after attached Flow¶
You can disable this setting and stop your main flow after an attached flow has been executed.
Continue Flow after default reply¶
Enable this setting to continue executing the flow logic after a default reply has been delivered.
Continue Flow after negative confirmation¶
Enable this setting to continue the flow logic after an intent confirmation sentence has been responded to negatively.
Pass Default Reply into Flow¶
If the setting is on, the virtual agent doesn't answer with the default reply but writes it in the input object. Afterward, this information can be used in order to do something with it in the flow logic.
Include Default Reply in NLU Training¶
If the setting in on, NLU training will also include default replies configured for each Intent as example sentence to match the intent. This setting can be also configured individually on intent level.
Set the timezone that your virtual agent will operate in.
Enable Case Sensitive Intent Mapping¶
When enabled, case sensitivity and punctuation will be taken into account by the intent mapper. Disable this setting to ignore casing and punctuation in ML intent mapping.
Despite configurable in the agent's settings, this feature is not implemented for the regular English locales, i.e. US, Indian, UK, Canadian, and Australian English. However, you may use this feature using the "Universal" locale which also supports general English.
Attached Flow Intent Mapping Priority¶
- Jointly Map Main and Attached Flow: treats Main and Attached Flow Intents as if they were a single intent collection and finds the best matching intent and executes the corresponding Flow.
- Map Main Flow first: first maps the Main Flow Intents separately. If no Intent is found in the Main Flow the Attached Flow is mapped.
- Map Attached Flow first: first maps the Attached Flow Intents separately. If no Intent is found in the Attached Flow the Main Flow is mapped.
Implicit Slot Parsing¶
Determines whether slots in NLU example sentences should be parsed implicitly or only slot annotations should be used for training. Configurable from 4.1.4 onwards. Prior to 4.1.4 the behavior was as in "Full" mode, existing agents will retain this default.
- Disabled: Only use slot annotations: This will only consider user-annotated slots, Lexicon, or Any Slots, in example sentences.
- Full: Parse both System and Lexicon slots implicitly
- System Slots - No Lexicon Slots: Parse only System slots like numbers, dates, email implicitly. Only consider annotations for Lexicon Slots.
- Lexicon Slots - No System Slots: Parse only Lexicon slots like from attached Lexicons implicitly. Only consider annotations for System Slots.
Forget question threshold¶
Number of contact responses after which a user's answer to a question is no longer registered.
Score from which on an Intent is considered confirmed if a confirmation sentence is set. Value between 0 and 1.
Score from which on an Intent is considered confirmed or marked for reconfirmation if a reconfirmation sentence is set. Value between 0 and 1.
Adjust the Reconfirmation Threshold
The confidence threshold has no effect unless the intent uses confirmation sentences. The Reconfirmation Threshold is your lower confidence bound - you must set it in addition to the Confidence Threshold. Intent scores above the reconfirmation threshold are confirmed or marked for reconfirmation.
If disabled, analytics data will not be collected and stored for the agent.
Allow data only¶
If disabled, analytics data will not be collected for messages with only a JSON data payload and no text input.