Relevance AI Setup Guide: Build Your First Agent Team

Relevance AI is the most powerful platform for building custom multi-agent workflows. This guide walks through building your first agent team — from initial setup to production deployment. Expect 1-2 weeks from setup to your first production workflow.

Step 1: Sign up and choose a plan

Sign up at relevance.ai. Start with the Free tier (100 runs/month) for evaluation. Upgrade to Pro ($400/month) when you're ready for production.

Step 2: Understand the architecture

Relevance AI uses a multi-agent architecture:

• Agents: Specialized AI workers with defined roles
• Tools: Functions agents can call (APIs, databases, etc.)
• Workflows: Multi-step processes that chain agents together
• Handoffs: How agents pass work to each other

Step 3: Build your first agent

Start with a simple single-agent workflow:

1. Create a new agent
2. Define its role: "Research target companies matching ICP criteria"
3. Add tools: web search, Apollo API, LinkedIn API
4. Configure the prompt with ICP criteria
5. Test with sample inputs

Step 4: Build a multi-agent workflow

Once your first agent works, chain it with others:

1. Agent 1 (Researcher): Find target companies
2. Agent 2 (Enricher): Find decision-maker contacts
3. Agent 3 (Writer): Draft personalized outreach
4. Agent 4 (Sender): Execute email sequence

Configure handoffs — Agent 1's output becomes Agent 2's input, etc.

Step 5: Configure safety

Before going live:

• Spending caps: Set monthly limits in billing settings
• Alerts: Configure alerts at 50% and 90% of cap
• Approval steps: Add human approval for external communications
• Audit logging: Enable for all agent actions

Step 6: Test and deploy

Test in sandbox mode with sample data for 1-2 weeks. Once confident, deploy to production with close monitoring for the first month.

Common mistakes to avoid

• Building too complex a workflow initially
• Not setting spending caps (a runaway agent can rack up significant charges)
• Skipping sandbox testing
• Not monitoring agent performance post-deployment

Next steps

Once your first workflow is reliable, add more. See our full Relevance AI review for capabilities, and our Lindy vs Relevance comparison for when to use each.