Skip to content

Quickstart

Deploy your first agent in 5 minutes using the RunAgents console. No code to write, no configuration to manage — the platform provides a sample agent and built-in tools to get you started instantly.

If you want a more production-shaped starting point, use the Agent Catalog after this quickstart. The catalog includes richer examples such as the Google Workspace assistant.

Prerequisites

You need a RunAgents account. If you do not have one yet, sign up at try.runagents.io.

Step 1: Log In to the Console

Open the RunAgents console at your platform URL and log in with your credentials.

When the platform is new or empty, the Dashboard displays a welcome screen with a prominent call to action.

Step 2: Deploy the Hello World Agent

Click "Deploy Hello World Agent" on the dashboard hero card.

Alternatively, navigate to Agents in the sidebar and click "+ New Agent".

The platform automatically seeds two starter resources for you:

Resource Description
Echo Tool A built-in tool that echoes back any message you send it. No external API keys needed.
Playground LLM A pre-configured LLM provider (OpenAI gpt-4o-mini) for testing.

Note

The starter resources are created automatically the first time you deploy the Hello World agent. They are labeled as demo resources and can be replaced with real services later.

Step 3: Review the Deploy Wizard

The deploy wizard has three steps:

Upload

The sample agent code loads automatically. It is a simple agent that:

  • Calls the Echo Tool to send and receive messages
  • Uses the Playground LLM to generate responses

The platform analyzes the code in real time and detects the tools and LLM providers it uses.

Wire

The wiring step shows what the analysis found:

  • Detected tools are matched to registered tools on the platform (Echo Tool)
  • Detected LLM usage is matched to available model providers (Playground LLM)
  • Agent name is pre-filled as hello-world

For the Hello World agent, everything is pre-wired. All indicators should show green.

Deploy

Review the summary and click Deploy. The platform:

  1. Creates the agent with its configuration
  2. Sets up a service account for the agent
  3. Binds any selected policies to the agent service account
  4. Starts the agent

Step 4: View Your Agent

After deployment, you are redirected to the Agent Detail page. Here you can see:

  • Status: The agent transitions from Pending to Running
  • Overview: Configuration, required tools, and LLM settings
  • Runs: A timeline of all invocations once the agent starts receiving requests

What Just Happened?

Here is what RunAgents did behind the scenes when you clicked Deploy:

  1. Code analysis -- The platform scanned the sample code using AST parsing and pattern detection. It identified outbound HTTP calls to the Echo Tool and LLM gateway usage.

  2. Tool wiring -- The detected tool calls were matched to the registered Echo Tool. The agent was configured with the correct tool URLs as environment variables.

  3. LLM wiring -- The detected model usage was matched to the Playground LLM provider. The agent was configured with the LLM gateway URL and model settings.

  4. Policy wiring -- The deploy flow links selected policies to the hello-world agent so outbound tool calls are authorized by policy.

  5. Agent deployment -- The agent was deployed with its service account, environment configuration, and networking rules. All outbound traffic from the agent flows through the platform's security layer.

The agent is now live. Any request it makes to the Echo Tool is automatically:

  • Checked against the access policy
  • Enriched with the correct authentication credentials
  • Forwarded with the end-user's identity

Next Steps

Now that you have a running agent, explore further:

  • Register your own tools -- Add external APIs and SaaS services that your agents need to call. See Registering Tools.
  • Deploy your own agent -- Upload your own code instead of the sample. The analysis engine supports Python agents with LangChain, LangGraph, CrewAI, and more.
  • Deploy from the agent catalog -- Start from a production-style blueprint such as the Google Workspace assistant. See Agent Catalog.
  • Set up identity providers -- Configure JWT-based authentication so your client applications can call agents with user identity. See Identity Providers.
  • Configure approvals -- Add approval_required policy rules and approver groups for sensitive operations. See Approvals.

Want to use the API or CLI instead?

You can do everything shown above programmatically.