Evose

SaaS 5-Minute Walkthrough

Sign up → create workspace → run your first Agent · End-to-end in 5 minutes

The fastest path: from zero account to a working Agent. About 5 minutes.

What You'll Accomplish

  1. Create an Evose Organization
  2. Enter the default Workspace
  3. Build a customer-service Agent from a template
  4. Chat with it on the Workbench

Prerequisites

  • An email address (for sign-up)
  • A device with a modern browser (latest Chrome / Edge / Safari)

Steps

1 · Sign up and create an Organization (1 minute)

  1. Visit https://app.evose.ai
  2. Sign up with email and verify
  3. Follow the wizard to create your Organization — the top-level container in SaaS multi-tenancy
  4. The system automatically creates a default Workspace

Organization vs Workspace

  • Organization = your company/team, the top-level container, manages models / tools / members
  • Workspace = isolated build sandbox; one organization can hold N workspaces

2 · Enter the default Workspace (20 seconds)

After sign-up, you land on the Workbench. Click "Workspace" in the bottom left to switch into the default workspace.

3 · Create your first Agent from a template (2 minutes)

Inside the workspace:

  1. Left menu → Apps · Agent
  2. Top right → New → choose template "Customer Service"
  3. The template prefills:
    • Role prompt: You are a friendly customer service rep — answer concisely and professionally
    • Base model: platform default LLM
    • Tools: empty (you can add later)
  4. Click SavePublish to Workbench

Which model is used?

SaaS uses platform-curated LLMs by default, billed in Credits. Under Private deployment you must configure your own model.

4 · Chat on the Workbench (1 minute)

  1. Top menu → back to Workbench
  2. In the Task menu, find the "Customer Service" agent you just published
  3. Click into the chat panel and try a message:

    I ordered a laptop — when will it ship?

  4. The Agent should reply in a customer-service tone and ask for the order number

5 · Take a peek at observability (30 seconds)

Back in the workspace → Workspace management · Observability — the conversation has been recorded:

  • Logs: dialog content
  • Metrics: token usage, response time
  • Traces: Agent's internal execution steps

Deep dive on Observability

Verification Checklist

After running through, you should have:

  • At least one published Agent visible on the Workbench
  • Completed at least one conversation with the Agent
  • Corresponding records in the observability panel

FAQ

Agent gives unreliable answers?

The default template is a generic rep — no knowledge base attached. Next, learn to configure a knowledge base and feed it your enterprise documents.

Want a stronger model?

Organization administrators can configure the default model on the Model platform; builders can pick a specific model on the Agent configuration page.

Want to share with multiple people?

Invite members → configure resource-level ACL on the Agent.

Next Steps