Automation Recipe: Incident Notification and Documentation Workflow

What This Builds

You'll create an automated workflow: when a worker submits an incident report form (via Google Forms, Microsoft Forms, or similar), the automation immediately notifies the right people, sends the reporter a confirmation, and uses ChatGPT to draft a preliminary investigation summary that lands in your inbox ready to review — all before you've even looked at your phone.

Instead of manually notifying your supervisor, HR, and the site manager for every incident and then starting your investigation from scratch, you arrive at the scene with the notification chain already in motion and a preliminary report draft waiting.

Prerequisites

• Zapier account (free tier handles this automation; paid tier for multi-step automations)
• A form tool: Google Forms (free), Microsoft Forms (included in Microsoft 365), or JotForm
• ChatGPT account (free tier works)
• Email addresses for the notification chain (supervisor, HR, site manager)
• 1–2 hours to build and test

The Concept

A Zapier automation is like a relay race: the first baton (form submission) triggers the next runner (email notification), who triggers the next (ChatGPT drafts the report), who triggers the next (your email). You set up the race once, and it runs automatically every time. You focus on the investigation itself — the administrative relay runs without you.

Build It Step by Step

Part 1: Create Your Incident Report Form

Build a Google Form with these fields:

• Reporter name and contact information
• Date, time, and location of incident
• Description of what happened (paragraph text — tell them to write in their own words)
• Was anyone injured? (yes/no)
• If yes: nature of injury, body part, medical treatment received
• Equipment or chemicals involved (if any)
• Witnesses (names/contact)
• Immediate actions taken

Set the form so responses are logged in a Google Sheet (in Google Forms: Responses tab → Link to Sheets). This becomes your trigger for the Zapier automation.

Part 2: Set Up the Zapier Automation

1. Sign in to zapier.com (or create a free account)
2. Click Create Zap
3. Trigger: Choose "Google Sheets" → "New Row" → connect to the spreadsheet that captures your form responses
4. Click Continue and test the trigger with a sample form submission

Part 3: Add Email Notification Step

1. Click + to add an action → Choose "Email by Zapier" (free) or connect your actual email provider (Gmail, Outlook)
2. To: Paste in the email addresses for your notification chain: supervisor, HR, site manager — separate with commas
3. Subject: Build dynamically: "EHS Incident Report — [map to date field] — [map to injury yes/no field]"
4. Body: Build a notification using the form fields:

A new incident report has been submitted:

Reporter: [map to reporter name field]
Date/Time: [map to date/time field]
Location: [map to location field]
Injury?: [map to injury yes/no field]
Description: [map to description field]

This is an automated notification. EHS Specialist [YOUR NAME] has been notified and will follow up within [your response time commitment].

Part 4: Add ChatGPT Draft Step

1. Add another action → Search for "ChatGPT" → Select "Send Message"
2. Connect your OpenAI account (requires an API key — free tier is enough for occasional use)
3. Model: Select {{tool:ChatGPT.model:general}}
4. Message: Build this prompt dynamically using the form fields:

You are an EHS specialist assistant. A worker has submitted an incident report. Based on the information below, draft a preliminary incident investigation summary that the EHS Specialist can review and expand.

Incident Details:
Reporter: [map to reporter field]
Date/Time: [map to date/time field]
Location: [map to location field]
Injury: [map to injury field]
Description: [map to description field]
Witnesses: [map to witness field]
Immediate Actions Taken: [map to immediate actions field]

Write:
1. A 150-word preliminary incident narrative (what appears to have happened)
2. Three initial investigation questions that should be asked when interviewing the reporter
3. A preliminary list of 3 potential root causes to investigate
4. An initial OSHA recordability assessment (yes/no/needs more info)

1. Save the ChatGPT response — Zapier captures the output text

Part 5: Email the Draft to Yourself

1. Add a final action → Email → send to your own address
2. Subject: "DRAFT Investigation Summary — [incident date] — [location]"
3. Body: Include both the original form submission AND the ChatGPT draft:

NEW INCIDENT REPORT — PRELIMINARY INVESTIGATION SUMMARY

--- Original Report ---
[map all form fields here]

--- AI-Generated Preliminary Summary (Review Before Using) ---
[map the ChatGPT response output]

Note: This is a preliminary AI draft. Verify all details before issuing the official investigation report.

Part 6: Test the Full Workflow

Submit a test incident form with fictional data. Verify:

• Notification email arrives within 2 minutes
• Your investigation draft email arrives with the preliminary summary
• The ChatGPT draft looks reasonable and reflects the test incident details

If the ChatGPT output is generic or off-topic, refine the prompt in the Zapier step to be more specific about your facility/industry.

Real Example: Incident Automation in Action

Setup: A warehouse EHS Specialist builds this Zap. Workers report incidents through a QR code-linked Google Form posted in the break room.

Input: A worker submits a report at 6:45 AM about a forklift near-miss in aisle 7 during the night shift.

What happens automatically (within 3 minutes):

1. Notification email goes to: Operations Manager, HR Director, Night Shift Supervisor
2. EHS Specialist receives a preliminary investigation summary in their inbox before 7 AM

The draft summary contains:

• "At approximately 5:30 AM, a near-miss occurred in aisle 7 involving a forklift and a pedestrian worker. The forklift operator was traveling at normal speed when the pedestrian stepped from an intersecting aisle without checking for traffic..."
• Interview questions: "Was the pedestrian in a designated walking aisle? What visibility conditions existed at the intersection? Was the forklift operator aware of workers in the area?"
• Root cause candidates: "Aisle intersection markings inadequate; mirror placement at blind intersection; pedestrian traffic patterns not separated from forklift routes"

Time saved: 45 minutes of manual notifications and investigation setup reduced to reviewing and refining a draft.

What to Do When It Breaks

• Zap doesn't trigger → Check that the Google Sheet is the correct one linked to your form; Zapier needs at least one row already in the sheet to set up
• Email doesn't send → Verify email addresses are correct and test the email step independently
• ChatGPT output is too generic → Add industry context to the prompt: "This is a warehouse/manufacturing/construction facility with forklift operations and [other specific context]"
• API key errors → Make sure your OpenAI API key has credit loaded (even small amounts work for occasional use)

Variations

• Simpler version: Skip the ChatGPT step and just build the automatic notification chain — even that saves 15–20 minutes per incident
• Extended version: Add a step that creates a task in your project management system (Asana, Monday.com, Jira) so corrective actions automatically get tracked

What to Do Next

• This week: Build and test the form and basic notification step (Steps 1–3)
• This month: Add the ChatGPT draft step and refine the prompt based on real incidents
• Advanced: Build a second automation for corrective action follow-up — when a corrective action is past its due date in your tracking sheet, automatically email the owner a reminder

Advanced guide for EHS Specialist professionals. These techniques use more sophisticated AI features that may require paid subscriptions.