Skip to main content

Getting Started

You're 10 minutes away from a working AI Agent. This page covers signup and the bare minimum to ship one. Once it works end-to-end, deeper customisation lives in the rest of the docs.

1. Create your account​

  1. Open the Cubeo AI sign-up page.
  2. Enter your email and a password — or click Sign up with Google to skip the password step.
  3. Optionally name your team. If you don't, Cubeo derives one from your email; you can rename it later.
  4. Log in.

2. Find your way around​

After login you land on the dashboard. The left sidebar exposes everything you'll use day to day:

  • AI Agents — every AI Agent in this team. Create, edit, delete from here.
  • Knowledge Bases — files, websites, Q&A pairs your AI Agents can read from.
  • Integrations — external accounts and MCP servers, in one place.
  • Dashboard — what's happening across every AI Agent right now.
  • Tasks Management — every task across every AI Agent, with filters.
  • Analytics — credits usage and AI Agent activity over time.

Spend a minute clicking around. You won't break anything.

3. Build your first AI Agent​

  1. Click Create new AI Agent on the AI Agents page.
  2. Name the AI Agent — short and descriptive (e.g. Support Agent).
  3. Description — one sentence of what it's for. Keeps your team aligned.
  4. Model — pick an LLM. GPT-4 and Claude Sonnet are good defaults.
  5. Visibility — leave it Private for now. You can flip it to Public later when you're ready to embed it on your site.
  6. Instructions — describe how the AI Agent should behave in plain English. Tone, what it should refuse, when to ask for clarification.
  7. Save.

Now your AI Agent exists, but it knows nothing and can do nothing. Two quick wins:

4. Test it before you ship​

Open your AI Agent's Chat tab and run a few questions. If the answers aren't right, tweak the instructions or add more knowledge — that's the loop you'll spend most of your time in.

Stay in control of sensitive actions

Turn on Human-in-the-Loop so your AI Agent pauses and asks for your approval before it writes to external systems (HubSpot, your API, etc.).

What's next​

  • How to create an AI Agent — deeper walkthrough of every option in the AI Agent form.
  • Knowledge Bases — teach your AI Agent from your files, websites, and Q&As.
  • Tools — the full catalogue of actions your AI Agent can take.
  • Integrations — connect HubSpot, Tavily, MCP servers, and more, once for the whole team.