Conversational Intelligence

Automatic Intent Disambiguation

Unlike search engines that just return anything that might be relevant, chatbots provide one response out of many possibilities. Most of the time it is clear what the intent is. Some utterances can be vague and yield multiple candidate intents. Pez.AI chatbots automatically detect this siuation and ask for clarification and displays a short list of potential intents. This is better than guessing, which can lead to incorrect responses.

Automatic Handling of Tangents and Non-Sequitors

Within a workflow, the bot expects the conversation to follow a specific sequence. Sometimes, people want to ask for more detail about a question. In a human conversation, we can naturally handle the tangent and then return to the original conversation. Pez.AI chatbots can do the same thing, where the bot will answer a tangential question and then return to the workflow.

Adaptive Known Unknowns

Unlike search engines that always give an answer, Pez.AI chatbots know what they don’t know. In these situations the bot escalates to a human. This model is dynamic and adapts to your corpus as it changes.

Adaptive Multilingual Support

Our chatbots detect multiple languages. If your bot has multilingual support enabled, you can create interactions in different languages. The bot will detect the language of the user’s utterance and respond in the same language. That means a single conversation can flow between different languages seamlessly.

Probe For Details

Our bots know when responses don’t have sufficient details. This is useful in interviews (Interviewer) as well as project management (Proctor). In these situations, a human would naturally ask for more detail. Our bots do the same thing, which ensures people get the level of detail they need to make decisions.