Skip to content

How Nuvoca works

A phone number routes a call to an Agent or an Agentic Flow. The selected agent listens, uses its instructions and context, and speaks a response. Tools connect the conversation to actions outside Nuvoca.

The call path

  1. The call reaches the Agent assigned to the phone number.
  2. Speech recognition captures what the caller says.
  3. The Agent responds using your instructions and the available information.
  4. The caller hears the response in your chosen voice.
Component What you configure Example
Routing Which agent answers and when Reception in business hours, service agent at night
Speech recognition How caller speech becomes text Main language and specialist vocabulary
Agent instructions Task, boundaries and conversation rules Ask for missing details one at a time
Model How the agent processes context and chooses actions An available model suited to the task
Voice How the answer is spoken Voice, delivery and pronunciation
Knowledge Base Approved company information Service coverage and product instructions
Tool A specific external action or lookup Fetch an order status or send a message

Context before, during and after the call

  • Pre Processing can look up caller details before the conversation. This lets the Agent greet a known customer by name, for example.
  • During the call, the Agent uses earlier details and relevant company information. Tools let it look up an order status or send a message.
  • Post Processing controls the final summary and its delivery by email or webhook. Profiles can apply different configurations to different agents or flows.

Set up the actions you need under Tools and integrations, then describe in the prompt when the Agent should use them.

Agents and Agentic Flows

A single Agent owns one conversation task. An Agentic Flow selects among specialized agents using intent routing. To connect the caller to a person on your team, use phone forwarding.

Three different languages

The interface language controls Nuvoca's menus. In speech recognition, choose the languages your callers use. Prompt and voice settings influence what the caller hears. Changing one does not automatically configure the others.

Next: create your first agent, or use the glossary to look up a term.