Getting started

Your first two weeks: from sources connected to agent live

Most teams expect a long, complex rollout before an AI agent can handle customer questions. In practice, the first two weeks of the Unless implementation timeline are about connection and training, not coding. By the end of week two, your sources are plugged in, your knowledge base is being built, and you are already running quality checks. The remaining two weeks focus on review and deployment, but the foundation is set early.

Week 1: Connect

The first week is dedicated to connecting your existing systems and configuring the initial settings. This is where you install the Unless snippet, set up your workspace, and add your first Source. Most customers have their first agent live within four weeks, and this week is the groundwork.

Install the snippet

The Unless snippet loads the Customer Agent on your site. You need admin access to your Unless account and the ability to add a script tag to the head of your website. If you use a tag manager such as Google Tag Manager or Segment, that works too. You do not need developer access to the Unless codebase.

To install the snippet:

  1. Open Account and billing > Settings and click Get Unless snippet
  2. Copy the script tag
  3. Paste it into the <head> of every page that should host the Customer Agent

The snippet loads asynchronously and does nothing visible until you publish a component or starter button, so installing it on a production site is safe. If you use a tag manager, add the snippet as a custom HTML tag that fires on all pages.

Configure your workspace

A workspace holds your entire Unless configuration. Most companies use one production workspace. You must configure three language settings:

To configure these, open Account and billing > Account settings.

Add your first source

The agent needs at least one source before it can respond to customer questions. Choose the source that answers the most customer questions today. This is usually your help center, your main marketing site, or your product documentation.

  1. Go to Train > Knowledge base > Sources
  2. Add the URL of your chosen source
  3. Unless will crawl it, process it into the Content library, and make it available to the agent

During this week, you also plug in your knowledge sources, helpdesk, and CRM, configure audiences, and set up compliance settings.

Week 2: Train

The second week focuses on building the knowledge base and running quality checks. This is where the Train phase of the Unless loop comes into play, turning your existing content into Living Knowledge.

Build the knowledge base

Your sources point at the documents, products, and pages the agent should learn from. The Content library captures the structured knowledge that emerges from ingestion, showing content nodes generated from your sources. Each source displays its connection status: Connected, Syncing, Error, Disconnected, or Team only.

Run test scenarios

Before a customer sees the agent, you can preview every role, simulate every skill, and audit every decision. Control questions and real conversations are replayed with audit trails your DPO and auditor can read. The Test phase works like this:

  1. Preview the role - See what the agent will say, in every language, before you ship it.
  2. Simulate the skill - Run agentic skills against synthetic users to confirm they behave as defined.
  3. Audit the decision - Per-interaction trail with timestamp, decision path, model used, and source citation.

Review quality reports

Quality reports track test scores over time with pass rates and trend indicators. Each report shows what passed and what needs attention. Reviewing these reports shows coverage gaps before going live. Every answer points back to its source, so you can read the decision behind any interaction in seconds.

What happens next

Once the snippet is installed, the workspace is configured, and a source is connected, the Engine starts building Living Knowledge. You can move on to the content library to see how your sources are structured, procedures to set up the actions the agent can take, audiences to segment your visitors, and preview to see the agent before customers do.

Week 3 focuses on stakeholder walkthroughs, fine-tuning responses, approving actions, and setting supervision levels. Week 4 moves the AI into production with initial customers, where the agent handles real traffic and every interaction feeds into the next training cycle.

Conclusion

The first two weeks are the foundation of your Unless deployment. By the end of week two, you have connected your systems, built a knowledge base, and run quality checks that show where the agent needs work. The remaining two weeks are about refinement and approval, but the heavy lifting is done. If you follow the four-week timeline, you can have a live agent on your own data within a month.