Creating an Agent

Step-by-step guide to creating your first Verow AI agent, from naming to configuration.

Creating an agent takes just a few minutes. This guide walks you through each step with best practices for getting the most out of your agents.

Step 1: Open the Agents Page

From your workspace, click Agents in the left sidebar. You'll see a list of any existing agents, plus a button to create a new one.

Click Create Agent to open the creation form.

Step 2: Choose a Name

Your agent's name should make it immediately clear what the agent is for. Team members will see this name when selecting which agent to chat with.

Naming Best Practices

For client agents:

  • Use the client's actual name: "Acme Corp", "Bright Horizons", "TechStart Inc."
  • Keep it simple and recognizable to your team

For project agents:

  • Include the project or campaign name: "Q2 Product Launch", "Website Redesign 2026"
  • Add the client name if it helps: "Acme - Holiday Campaign"

For department agents:

  • Describe the function: "RFP Assistant", "Creative Brief Helper", "Onboarding Guide"
  • Be specific about what it does

Names to Avoid

  • Generic names like "Agent 1" or "Test"
  • Abbreviations that not everyone knows
  • Names that could apply to multiple things

Step 3: Write a Description

The description tells the agent what it should know and how it should behave. Think of it as instructions for a new team member joining the account.

A good description includes:

Who the agent serves: "This agent supports the Acme Corp account."

Key context: "Acme is a B2B software company focused on enterprise customers. They value data-driven recommendations and prefer formal communication."

Behavioral guidance: "Always reference past campaign performance when making recommendations. Flag any requests that might exceed their quarterly budget."

Example Descriptions

Client Agent:

This agent supports the Acme Corp account.

Key information:
- B2B enterprise software company, founded 2018
- Primary contact: Sarah Chen (VP Marketing)
- Preferred communication style: Data-driven, concise
- Brand voice: Professional, innovative, trustworthy
- Current retainer: $15,000/month for ongoing content and campaigns

When responding:
- Reference past campaign metrics when relevant
- Flag any scope changes that might affect budget
- Maintain formal but friendly tone

RFP Assistant:

This agent helps the sales team respond to RFPs and proposals.

Access to:
- Past winning proposals and case studies
- Standard pricing and scope templates
- Client testimonials organized by industry

When drafting responses:
- Match the tone and requirements of the specific RFP
- Pull relevant case studies from similar industries
- Flag any unusual requirements for human review
- Include clear next steps and timelines

Step 4: Configure Agent Context

After creating the basic agent, you can connect it to specific data sources and context. This tells the agent what information it should draw from when answering questions.

Context options include:

  • Files: Upload relevant documents (brand guidelines, contracts, briefs)
  • Links: Add web resources your agent should reference
  • Meetings: Connect your calendar to capture client conversations automatically

The more relevant context you provide, the better your agent's responses will be.

Step 5: Save and Test

Click Create Agent to save your new agent. You'll be taken to the agent's chat interface where you can immediately start a conversation.

Test Your Agent

Before sharing with your team, test the agent with a few questions:

  1. Basic recall: "What do you know about this client?"
  2. Specific details: "What's their preferred communication style?"
  3. Practical task: "Draft a brief status update email for this week."

If responses seem off, go back and refine the description or add more context.

Quick Reference: Agent Creation Checklist

  • [ ] Choose a clear, recognizable name
  • [ ] Write a description that includes key context and behavioral guidance
  • [ ] Connect relevant documents and projects
  • [ ] Test with 2-3 questions before sharing with team
  • [ ] Share the agent with team members who need access

Next Steps

Your agent is ready to use. Learn how to get the most out of conversations in the next guide.