Skip to content
Omni Octo Docs

Overview

Omni Octo (“Octo”) is the AI agent behind your workspace. It has one brain — the same tool catalog, role filtering, confirmation logic, and conversation memory — exposed through three different surfaces depending on how you want to integrate.

The pitch

Section titled “The pitch”

If your tool already speaks the OpenAI API, you can talk to Octo by changing two things: base_url and api_key.

from openai import OpenAI


client = OpenAI(
    api_key="sk-octo_...",  # your Octo API key
    base_url="https://api.omniocto.com/v1",  # flip this
)


resp = client.responses.create(
    model="claude-sonnet-4-6",
    input="What can you help me with?",
)
print(resp.output_text)

Every request — however it arrives — runs through the same internal turn pipeline that powers the dashboard and the email/chat channels, so you get Octo’s full tool catalog, role-based permissions, spend confirmations, and usage metering unchanged.

Three surfaces

Section titled “Three surfaces”
SurfaceWhat it’s forStart here
OpenAI-compatible API (/v1)Drop-in replacement for OpenAI SDKs, LangChain, LiteLLM, n8n, and any tool with a “custom OpenAI-compatible endpoint” field. Best for chat/agent integrations.OpenAI-compatible API overview
MCP serversConnect an MCP-native client (Claude Code, Claude Desktop, other MCP hosts) directly to Octo’s tools or to a specific workspace conversation.MCP overview
Platform REST API (/api/*)The full dashboard API — agents, campaigns, contacts, conversations, voice, billing. For advanced integrations that need direct resource access rather than a conversational interface.Platform API overview

Before you start

Section titled “Before you start”

All three surfaces authenticate with the same kind of credential: a workspace API key. See Get an API key to mint one from the dashboard, then Authentication for how the key maps to your permissions.

Where to go next

Section titled “Where to go next”