Getting Started with Relay: A Complete Onboarding Guide

Getting up and running with Relay takes less time than most teams expect. The platform is designed so you can connect your inbox, add your knowledge base, and start seeing AI-drafted replies within a single session. This guide walks through every step of the setup process, from creating your account to sending your first AI-assisted response.

What You Will Need

Before starting setup, make sure you have:

• A Gmail or Microsoft Outlook email account that you want to connect for support
• Access to your knowledge base content (help articles, FAQs, product documentation, or similar reference material)
• About 30 minutes for initial setup

That is it. Relay handles the rest of the infrastructure, including AI model configuration, email integration, and knowledge base indexing.

Step 1: Create Your Account and Workspace

When you sign up for Relay, the first thing you create is a workspace. A workspace is your team's shared environment where mailboxes, agents, and knowledge bases live. Think of it as your support command center.

Choosing Your Workspace Name

Pick something that identifies your team or company. If you run support for Acme Corp, "Acme Support" works well. You can always change this later, so do not overthink it.

Inviting Team Members

You can invite team members during setup or later. Each team member gets a role:

• Owner: Full administrative access, including billing and workspace settings
• Admin: Can manage mailboxes, agents, and team settings, but not billing
• Agent: Can review and send AI-drafted responses, manage conversations, and contribute to the knowledge base

For initial setup, you only need one person. Invite the rest of your team once the system is configured and you are happy with how it works.

Step 2: Connect Your Email

Relay integrates directly with Gmail and Microsoft Outlook, the two email providers that handle the vast majority of business email. Connecting your email takes a few clicks through a standard OAuth authorization flow.

Connecting Gmail

1. Navigate to the Mailboxes section in your Relay dashboard
2. Click "Add Mailbox" and select Gmail
3. You will be redirected to Google's authorization screen
4. Sign in with the Gmail account you want to use for support
5. Grant the requested permissions (Relay needs access to read and send email on your behalf)
6. You will be redirected back to Relay with your mailbox connected

Connecting Microsoft Outlook

1. Navigate to the Mailboxes section in your Relay dashboard
2. Click "Add Mailbox" and select Microsoft Outlook
3. You will be redirected to Microsoft's authorization screen
4. Sign in with your Microsoft account
5. Grant the requested permissions
6. You will be redirected back to Relay with your mailbox connected

What Happens After Connection

Once your email is connected, Relay begins syncing your recent email history. This gives the AI context on your existing conversations and helps it understand the types of questions your customers ask. The sync runs in the background and typically completes within a few minutes.

Relay only accesses email in the connected mailbox. It does not access personal email accounts or other mailboxes unless you explicitly connect them.

Step 3: Build Your Knowledge Base

The knowledge base is what makes Relay's AI responses accurate and specific to your business. Without a knowledge base, the AI can only generate generic responses. With a good knowledge base, it drafts responses that sound like they came from someone who knows your product inside and out.

What to Include

Start with the content that covers your most common customer questions:

• Product FAQs: Answers to the questions you get asked most often
• How-to guides: Step-by-step instructions for common tasks
• Pricing and plan information: What each plan includes, how billing works, how to upgrade or downgrade
• Policy documentation: Refund policy, terms of service summaries, SLA commitments
• Troubleshooting guides: Common problems and their solutions

Adding Content

Relay supports several ways to add knowledge base content:

• Direct text entry: Paste or type content directly into the knowledge base editor
• Document upload: Upload existing documentation files
• URL import: Point Relay to existing help center pages to pull in content

Organizing Your Knowledge Base

You can create multiple knowledge base collections to organize content by topic, product, or audience. Each AI agent can be configured to reference specific collections, so you can customize which information is available for different mailboxes.

For most teams starting out, a single collection with all your support content works fine. You can reorganize into multiple collections later as your content grows.

Step 4: Configure Your AI Agent

Every mailbox in Relay has an AI agent that you can customize. The agent configuration controls how the AI drafts responses and what personality it brings to customer interactions.

Setting the Tone

Relay lets you guide the AI's communication style. You can set your agent's tone to match your brand:

• Professional and formal: Appropriate for enterprise, legal, or financial services support
• Friendly and conversational: Works well for consumer products and SaaS companies
• Technical and precise: Ideal for developer tools and technical products

You can also provide specific instructions, like "Always greet customers by their first name" or "Include a link to our help center when relevant."

Choosing Your AI Model

Relay supports multiple AI providers, giving you the flexibility to choose the model that works best for your use case:

• OpenAI (GPT models): Excellent general-purpose response quality
• Anthropic (Claude models): Strong at nuanced, thoughtful responses
• Google (Gemini models): Good balance of speed and quality

You can experiment with different models and switch at any time. The knowledge base and configuration stay the same regardless of which model you use.

Setting Response Guidelines

Beyond tone, you can give your AI agent specific guidelines for how it should handle different types of situations:

• How to respond when the answer is not in the knowledge base
• Whether to proactively offer additional help or keep responses focused
• How to handle sensitive topics like billing disputes or account cancellations
• Whether to include greetings, sign-offs, or specific formatting

Step 5: Choose Your Workflow Mode

One of the most important decisions in your setup is choosing between approval mode and auto-send mode. This controls what happens after the AI drafts a response.

Approval Mode (Recommended for Getting Started)

In approval mode, every AI-drafted response goes into a review queue. Your team sees the draft, the original customer email, and relevant knowledge base sources. Agents can:

• Approve and send: Send the draft as-is with one click
• Edit and send: Make changes to the draft before sending
• Discard and write manually: Start over with a manual response if the draft is not useful

Approval mode is the safest way to start because you see every response before it goes to a customer. It also gives you valuable data on how well the AI is performing and where your knowledge base has gaps.

Auto-Send Mode

Once you are confident in the AI's accuracy for certain types of questions, you can enable auto-send. In this mode, the AI sends responses automatically without human review.

Auto-send is best for:

• Routine questions with clear, factual answers
• Topics where your knowledge base is comprehensive and accurate
• Low-risk inquiries where an occasional imperfect response would not cause problems

Most teams start in approval mode and gradually enable auto-send for specific categories as they build confidence. This incremental approach lets you expand automation at a pace you are comfortable with.

Step 6: Process Your First Emails

With everything configured, it is time to see Relay in action.

Sending a Test Email

The easiest way to test your setup is to send an email to your connected mailbox from a different account. Ask a question that your knowledge base should be able to answer. Within moments, you should see an AI-drafted response appear in your Relay dashboard.

Reviewing the Draft

Open the draft and examine it:

• Does it answer the question accurately?
• Does it use information from your knowledge base?
• Does the tone match your brand?
• Would you feel comfortable sending this to a customer?

If the draft is good, approve and send it. If it needs work, edit it and send the improved version. Either way, this first interaction tells you a lot about where your setup is strong and where it might need tuning.

Iterating on Your Setup

Your first few days with Relay are a calibration period. Pay attention to:

• Which drafts need the most editing: This tells you where your knowledge base has gaps
• Which types of questions get the best drafts: These are candidates for eventual auto-send
• Tone mismatches: Adjust your agent configuration if the AI's communication style does not quite match your brand
• Missing information: Add content to your knowledge base whenever you notice the AI is missing key details

Tips for a Successful Launch

Start with One Mailbox

If you manage multiple support email addresses, start with just one. Get the workflow working well for a single mailbox before expanding. This lets you learn and adjust without multiplying complexity.

Brief Your Team

Before rolling out Relay to your support team, walk them through the workflow:

• How to find and review AI drafts
• How to approve, edit, or discard drafts
• How to flag knowledge base issues when they find them
• That the AI is a tool to help them, not a replacement for their judgment

Set Realistic Expectations

AI-drafted responses are not perfect out of the gate. They get better as you improve your knowledge base and refine your agent configuration. Set the expectation with your team that the first week is about calibration, not perfection.

Use the Feedback Loop

Every time an agent edits an AI draft, that is valuable feedback about where the system can improve. Review these edits regularly and use them to identify knowledge base updates, configuration changes, or training needs.

Next Steps After Setup

Once you are comfortable with the basics, explore these features to get more value from Relay:

• Team collaboration: Assign conversations to specific agents, add internal notes, and collaborate on complex issues
• Analytics: Review response time metrics, draft acceptance rates, and volume patterns
• Multiple mailboxes: Connect additional email accounts for different support channels or products
• Knowledge base expansion: Continuously add and refine content to improve AI accuracy
• Auto-send configuration: Gradually enable auto-send for high-confidence response categories

Relay is designed to grow with your team. The setup you build today is just the starting point. As you add content, refine your configuration, and learn from your team's interactions with the AI, the system becomes more capable and your support becomes more efficient.