The teams that set up Outlearn fastest are not the ones with the most technical experience. They are the ones who spent five minutes understanding how the pieces fit together before touching a single setting.

Outlearn is built around five concepts: your Agent, Sources, Actions, Handoffs, and Deployment. Each one has a clear job. Once you understand what each does and how they connect, every configuration decision becomes obvious - you will know immediately which lever to pull, which setting to change, and which part of the system to improve when you want better results. Spend five minutes here and the rest of the setup takes care of itself.

In this article, you'll learn:

• What the five core concepts are and what each one does
• How they connect to each other
• What a typical conversation looks like from start to finish

The Five Core Concepts

1. Your Agent

Your agent is the AI at the heart of Outlearn. It's what your users talk to. You give it a name, a personality, and a set of instructions - and it uses everything you connect to it to answer questions, take actions, and escalate when needed.

You can have more than one agent. A customer-facing support agent for your website and a separate internal agent for your team, for example. Each agent is fully independent - its own sources, its own instructions, its own analytics.

2. Sources

Sources are the content your agent learns from. This is where you connect your Helpjuice knowledge base, upload documents, add website URLs, or write short answers manually (called Snippets). When a user asks a question, your agent searches everything in your Sources to find the best answer.

The more relevant content you connect, the better your agent performs. An agent with rich sources answers confidently. An agent with no sources has nothing to say.

Think of Sources as your agent's memory.

3. Actions

Actions are tasks your agent can perform during a conversation - not just answer questions, but actually do things. Check a calendar for available slots. Send a confirmation email. Look up an order via an API. Search the web for up-to-date information. Actions turn your agent from a question-answering tool into something that can actually resolve problems end to end.

Actions are optional - you only need them if you want your agent to go beyond answering questions. But for teams that add them, it changes what the agent is capable of entirely.

Think of Actions as your agent's hands.

4. Handoffs

Even the best AI agent has limits. Handoffs let your agent recognise when a conversation needs a real person - and transfer it smoothly to your support team, along with the full conversation history. No repetition. No context lost. The human picks up exactly where the agent left off.

You can connect Outlearn to Intercom, Zendesk, or email, and define exactly when a handoff should happen - when the user asks to speak with a human, when the issue isn't resolved after a set number of attempts, or when a specific topic is raised.

Think of Handoffs as your agent knowing its limits.

5. Deployment

Deployment is where you decide where your agent lives. Embed it as a chat widget on your website. Connect it to your email inbox. Add it to Slack or Microsoft Teams. Install it as a Copilot inside Zendesk or Intercom so your agents get AI-drafted replies at their fingertips.

One agent can be deployed across multiple channels at the same time. The same knowledge, the same instructions, everywhere your customers are.

Think of Deployment as where your agent shows up for work.

How They Fit Together

Here's what a typical conversation looks like:

1. A user opens the chat widget on your website and asks a question.
2. The agent searches its Sources for the best answer and responds instantly.
3. If the answer requires live data - an order status, an available slot - the agent triggers an Action to look it up.
4. If the user asks to speak with a human, or the agent can't resolve the issue, a Handoff transfers the conversation to your support team.
5. All of this happens inside the Deployed channel - whether that's your website, your inbox, or your helpdesk.

Best Practices

• Set up your Sources before anything else - an agent with no content can't answer anything.
• Don't overcomplicate your first setup. Start with Sources and Deployment, then add Actions and Handoffs once your agent is live and working.
• Use the Preview panel throughout the dashboard to test how your agent responds before going live.