1. Quickstart: Your First Automation
About five minutes. Build a two-Step automation with a Scheduler and an AI Prompt Step. Run it manually and read the generated note in Draft.
- Use it to
- Build and run a two-Step automation (Scheduler + AI Prompt) in about 5 minutes
- Find it
- Automations → Create new → Workflow
- Cost
- 0 Step credits · ~320 Glow AI tokens per test run
- Output
- {{ 4.result }} (a short morning note in the Executions tab)
You need a Glow account and workspace with Step credits and Glow AI tokens. No external apps or code required. Check the Glossary for any unfamiliar terms.
What you will learn: Place a trigger and Step from the dock. Reference Steps by their fixed badge number, run a Draft manually, and read output in Executions.
Step 1: Create the automation
- Open App and select your target workspace.
- In the sidebar, click Automations → Create new → Workflow (or Create automation on an empty workspace). See Navigating the Dashboard if needed.
- On desktop, close the Workflow Assistant panel (X) to build manually. To generate drafts from a prompt instead, see Workflow Assistant.
The bottom bar (Apps, Tools, AI, Subflows, Search) is the dock. Everything you pick from the dock lands on the canvas.
Step 2: Add the trigger
Draft is not a sandbox. Keep this automation in Draft so the automatic schedule stays off. Manual runs still execute real actions and spend workspace allowance; see How testing works.
Open the trigger placeholder
Click the grey Click to Add Trigger box on the canvas.
Choose Scheduler
Select Tools → Start → Scheduler (or search Scheduler). It replaces the placeholder and opens its settings drawer on the right.
Look at the schedule
Leave Schedule mode set to AI Assistant and Schedule description set to Run everyday at 8:00 AM. Under Optional Props, Timezone defaults to your browser’s time zone.
- Describe when to run in plain English (for example, 'Run everyday at 8:00 AM').
- Check Next runs to verify the calculated run times before going Live.
- Enter a standard 5-field cron expression directly.
- Also previews upcoming execution times under Next runs.
Step 3: Add the AI step
Open the action placeholder
Close the Scheduler drawer (X), click Click to Add Action, and open the AI section.
Choose AI Prompt
Under Glow AI (No API keys required. Uses Glow AI tokens.), click AI Prompt. Skip the LLM models (API key required) provider list below it.
Write the instruction
Replace the sample Instructions text with:
Write a two-sentence motivational note to start the working day.
Step 4: Notice the step numbers
Check the number badge on each Step. A Step’s number is assigned when created and never reused, so numbers may not be 1 and 2.
Downstream Steps reference output using that exact badge number, such as {{ 4.result }}.
Always read the number off the Step badge. Referencing a non-existent Step number stops the run with “Placeholder 99.nothing didn’t find data”. You can also pick fields using the data icon beside any input (covered in Tutorial 2).
Step 5: Run it
Confirm the status badge next to the automation name says Draft, then click Run on the left side of the dock. Each Step displays a run counter badge once completed.
Step 6: Read what it wrote
Open the step
Click the AI Prompt Step on the canvas.
Switch to Executions
Select the Executions tab to view one row per run with its timestamp and execution ID.
Expand the run
The newest run opens expanded, showing token usage (for example, Glow-funded, not charged, 319 Glow AI tokens) and the result field:
result: Embrace this new day with renewed energy, knowing that your
unique skills and dedication make a real difference…
You have finished the tutorial once result displays your generated note. AI Prompt stores its text in result, so a later Step reads it as {{ 4.result }} (using your AI Step’s badge number). If a run fails, check Testing & Debugging.
What you just built
- Zero external setup: Built and ran in the browser with no API keys or connected accounts.
- Permanent Step numbers: Steps reference each other by badge number (
{{ 4.result }}), not canvas position. - Inspectable runs: Every Step records its output and token usage under Executions.
Optional: send the result on a schedule
1. Add a destination
Add a Slack or email Step after AI Prompt and connect your account. Pick a test destination and map {{ 4.result }} using your AI Step’s badge number.
- Run
- Keep the automation in Draft, click Run in the dock, then open the AI Prompt Step's Executions tab.
- You should see
- The latest run opens expanded and result contains a two-sentence motivational note.
What’s Next?
Try a quick prompt edit: Replace the AI Prompt Instructions with:
Write a two-sentence morning note that suggests one practical way to prioritize today's work. Avoid motivational slogans.Click Run and compare the new result in Executions.
👉 Continue to Tutorial 2: Build It Yourself → to pass data between Steps using AI Prompt or Slack.
Already comfortable mapping data? Branching adds region-based routes with Slack and Google Sheets.