What Is Expert?

Pez.AI Expert is a knowledge-sharing bot that starts from zero and builds up its own knowledge base over time by means of Dynamic Training.

The primary motivation for Expert is to help streamline information regarding processes and systems within organizations — knowledge which are usually siloed to a handful of persons. Having this problem makes organizations less productive as information sharing is time consuming and is often prone to inconsistent information being shared.


Pez.AI uses specific terminology to describe different human-bot interactions possible within the system.

  • Utterance - The input sentence or phrase expressed by a user to the bot.
  • Interaction - An utterance paired with a bot response, also referred to as a question-answer (QA) pair.
  • Conversation - A self-contained sequence of interactions.
  • Lexicon - The vocabulary or knowledge base of the bot. All interactions are recorded in the lexicon.
  • Intent - The underlying goal or intention driving the utterance. The intent can be obvious, such as “update my beneficiaries”. It can also be implied, such as “i’m sick”, which has the intent “take sick day”.

Pez.AI Expert also uses Expert-specific terms:

  • Topic - Topics are an abstract group of related concepts and may represent a discipline (e.g. marketing, data science), a technique (e.g. A/B testing, mathematical optimization), or a component in a system, or a process.
  • Handoff - The process of passing a question from one expert to another.

Key Features

Dynamic training

Dynamic training is a feature of Expert, where a Question and Answer pair gets saved in the lexicon after the question has been answered.

Normally, the process of adding an interaction would be to go to the Authoring > Train > Question Tab and manually add a QA pair. Dynamic training automates this process by immediately adding the QA pair to the lexicon once a question gets answered.

Question queueing


When an expert or a fallback receives multiple questions to answer, the questions are stored in a queue until answered. Once the question gets answered, it triggers dynamic training and it gets saved in the knowledge base.

Questions are assigned on a first come, first served basis.

Question handoff

Questions are assigned to a random expert or fallback. If they do not want to answer, they can hand off a question to someone. The bot directs the question to the new expert and waits for their answer (or the new expert can also hand off the question). Handofftrigger Handoffanswered

App integrations

Integrations Pez.AI Expert can be easily installed via dashboard to different applications for your specific needs. Expert can be integrated with Freshchat, Slack, or as a Webchat on your website.

We are open to adding channels! You may contact [email protected] for more information.


Pez.AI Expert has different pricing plans depending on your needs. They can be viewed here.