Skip to content

Introduction to 008

Welcome to the 008 Agent documentation, your framework for building voice-only, multilingual AI agents with multiple synthetic voices. In this “Getting Started” section, we’ll cover:

  1. What is a voice-first AI Agent?
  2. What are Tools? (built-in vs. third-party)
  3. What are Deployments?

A voice-only AI Agent in 008 Agent is a prompt-driven conversational entity that:

  • Speaks and listens via audio channels (no text UI).
  • Supports multiple languages out of the box.
  • Offers a variety of voices (male, female, accents, neutral, etc.).
  • Operates prompt-first, meaning all logic and dialog are defined through dynamic prompts rather than flowcharts or state machines.
  • Prompt-First: Behavior is entirely controlled by the LLM prompts you provide.
  • Multilingual: Specify the language in your prompt (e.g., "Answer in Spanish").
  • Voice Selection: Choose from platform-provided voice models or custom recordings.

Tools are functions or APIs the agent can invoke during a call. There are two categories:

These are native to 008 Agent for managing calls (only for SIP deployments):

  • hangUp: End the current call.
  • call_transfer: Transfer the call to another number or agent.
  • call_hold: Puts the call on hold. Can only be used if it is explicitly required by other tools.
  • call_unhold: Unholds a held call.
  • call_info: Use it to get all the call info. Useful to get current time and caller number, expect CDR data.

Clients can integrate their own APIs or services as external tools:

  • CRM Lookups: Query customer data by ID.
  • Ticketing Systems: Create or update support tickets.
  • Database Queries: Retrieve or store call metadata.
  • Custom Webhooks: Trigger operations in your backend.

Define each third-party tool with a name, input/output schema, and endpoint URL, so the agent knows how to call it.

Deployments are the endpoints through which your voice agent connects:

  • Web (Web Voicebot): Embed a voice interface into web pages or applications, allowing live browser-based calls.
  • SIP: Connect agents as VoIP extensions using SIP, handling inbound and outbound telephone calls.

These deployments determine the audio transport layer:

  • Web Voicebot: Users speak via the browser’s microphone.
  • SIP: Agents join PSTN or VoIP networks as phone extensions.