Running your first agent

There are two ways to run an agent in Z2X: through your plan steps, or directly from the command panel on the Home screen.

Method 1: Run a plan step agent

This is the most common way to run agents. Each step in your Z2X Path has an assigned agent ready to execute.

1. Go to your idea → Validate (or the plan section)
2. Find a step with status Ready
3. Click the step to expand it
4. Review any required inputs — fill them in if prompted
5. Click Run agent

The agent starts immediately. You'll see a loading indicator while it runs. Most agents complete in 20–60 seconds, though some complex steps (like a full competitor analysis) may take up to 2 minutes.

Method 2: Run from the command panel

The command panel on the Home screen lets you direct agents conversationally.

1. Make sure you're in Agent mode (the mode switcher is in the top-right of the command panel)
2. Type a question or instruction, for example:
   • "Who are the top 5 competitors for my AI note-taking idea?"
   • "Draft a cold outreach email for design agency owners"
   • "What's the best channel to find my first 10 customers?"
3. Press Enter or click the arrow button

Z2X reads your active idea brief and routes the request to the most appropriate agent. The response appears in the panel below the command input.

Active idea context

Commands from the Home screen are run in the context of your active idea. Make sure the right idea is selected (shown in the top bar) before running a command.

What happens during a run

1. Your request is sent to the Cloudflare Worker (the Z2X backend)
2. Credits are deducted atomically (the system checks your balance first)
3. The agent runs with your brief, profile, and any prior context as inputs
4. Output is returned and displayed in the artifact panel

If the agent run fails (network error, etc.), no credits are consumed. If it succeeds, credits are deducted whether or not you approve the output — the agent did the work.

Running agents in Autopilot

Autopilot mode runs agents automatically without you having to manually trigger each one. See Autopilot for a full walkthrough.

After the agent runs

The output appears in the artifact panel (in plan step view) or in the command panel response area (in Home view). See Reading agent output →.