How to Use ChatGPT to Review SWMS
High-risk construction work in Australia can't start until the Safe Work Method Statement (SWMS) has been checked. If you're the principal contractor, that can mean a stack of subbie SWMS landing on your desk at once, and it's not always easy to spot the ones copied from a master. Paste or upload a SWMS into ChatGPT and it'll point out the gaps: hazards without controls, controls that don't fit the work, generic text and missing risk ratings. You decide what changes and you sign it off.
Trusted by companies, projects and teams of all sizes
The short answer
Paste or upload the SWMS into ChatGPT and ask for a review. It checks each job step against its hazards, makes sure every hazard has a control, and looks at the risk ratings and sign-on. It won't give you a pass or fail. Instead, you get a list of gaps for the competent person to work through before signing off.
How to Get ChatGPT Reviewing Your SWMS
How good the review is depends on what you ask. Give ChatGPT the SWMS along with your standard and ask for specific findings. Then save the prompt so you can reuse it on every subbie SWMS.
-
Step 1
Give ChatGPT the SWMS and Your Standard
Paste the SWMS into ChatGPT or upload the file, then give it your standard: the high-risk construction work it should cover, a control for every hazard, a risk rating at each step, and job steps that match the real work.
- The SWMS, as pasted text or an uploaded file
- The task and site it covers
- Your own SWMS template or the principal contractor's checklist, if you have one
- The points that matter most to you, such as controls, risk ratings and sign-on
-
Step 2
Ask for a List of Gaps, Not a Verdict
Ask ChatGPT whether a SWMS looks fine and it'll usually say yes. Ask for a list instead, with each finding tied to a job step, so you can send it straight back to the subbie.
- Hazards named with no control
- Controls with no matching job step
- Generic text that fits any job
- Steps with no risk rating, or no rating after controls
- No sign-on section or named approver
-
Step 3
Make the Call on Each Finding
Treat the list as a first pass. Check each finding against the site and task, then send the real gaps back to whoever wrote the SWMS. The competent person still signs off.
Prompt Guidance for Reviewing SWMS with ChatGPT
A specific prompt gets a sharp review. Name the high-risk construction work the SWMS should cover and ask for findings against a clear standard. Once a prompt works, save it so every SWMS gets checked the same way.
- "Review this SWMS for [task] at [site]. List every hazard with no control and every control that isn't linked to a job step."
- "Go through each job step in this SWMS and flag any step with a missing risk rating, or a rating that doesn't match the hazard."
- "Quote any lines in this SWMS that are generic enough to fit any job, so I can send them back to the subbie."
- "Compare this SWMS against the attached template and list anything the template asks for that's missing."
- "Check whether this SWMS names the high-risk construction work and an approver, and whether it has a sign-on section. Tell me what's missing."
- "Before I sign off, give me the top five gaps in this SWMS, most serious first."
Troubleshooting Thin Results When Reviewing SWMS
When findings come back thin, the SWMS template is usually the cause. ChatGPT can only review what the form records, so the fixes go into the template your subbies fill in.
- Controls don't line up with hazards: they sit in separate tables, so put each hazard and its control on the same row
- Risk ratings come back missing: the matrix is an image or scan, so add a text field for the rating before and after controls
- Every SWMS gets the same findings: the form has no job fields, so add the site and task, along with the high-risk construction work
- Sign-on never gets checked: it's attached as a scanned sheet, so build the names and dates into the form itself
- Standard controls get flagged as vague: the wording changes each time, so use a pick list
Limitations
ChatGPT gives you a gap list, and the competent person keeps the sign-off. It can also help you draft a SWMS, but it doesn't run the work. On site, the crew still signs on to the version they're working to.
What ChatGPT Does Well When Reviewing a SWMS
Give ChatGPT the SWMS and a clear standard to check against. It reads fast and flags the lines worth a closer look.
- Reads the whole SWMS against what a good one should include
- Flags hazards without controls
- Flags controls with no matching job step
- Spots generic text that doesn't fit the task
- Checks risk ratings before and after controls
- Compares it with your template or the principal contractor's checklist
What ChatGPT Leaves to You When Reviewing a SWMS
Some decisions have to stay with the people on the job, and a review never replaces the sign-off.
- Knowing the site and the ground conditions on the day
- Judging whether a control is adequate
- Signing off or confirming WHS duties are met
- Catching everything, so your own review still matters
What This Assumes You Already Do
This assumes you already collect a SWMS from each subbie before high-risk construction work starts, and that someone checks it. ChatGPT speeds that check up. It doesn't replace the person who owns it.
- You collect a SWMS from every subbie before high-risk construction work starts
- Someone on your side owns checking each SWMS
- A competent person signs off each SWMS or sends it back
- You know the work well enough to tell a site-specific SWMS from a generic one
Why Use ChatGPT to Review SWMS
Reviewing one SWMS properly takes time, and before a Monday start you might have a dozen waiting. Under that pressure, checks get rushed and a generic SWMS can slip through. If something goes wrong, the trail leads back to the missing control and whoever signed it. ChatGPT gives you a structured first pass, so you can focus on the calls that matter.
What a Good SWMS Review Catches
A good review checks the SWMS against the WHS regulations and what your principal contractor expects. In most states and territories, a SWMS has to name the high-risk construction work and its hazards, risks and controls, and say how those controls are put in place, monitored and reviewed. Most principal contractors also want a risk rating and a crew sign-on.
- Whether the high-risk construction work is named plainly
- Hazards that are named but have no control
- Controls with no matching job step, or that lean on PPE alone
- Generic text that would fit any job or site
- Risk ratings that are missing or not updated after controls
- No named approver or sign-on record
It's Sunday night and six subbie SWMS are sitting in your inbox for a Monday start. Your name goes on the sign-off, and a skim lets the generic ones through.
Connecting Sitemate MCP to ChatGPT
Connecting takes three steps, and the first happens outside ChatGPT: an Org Controller turns on MCP access in Dashpivot. You'll also need ChatGPT Plus, Team or Enterprise, because connectors aren't available on the free plan.
-
Step 1
Turn On MCP Access in Dashpivot
An Org Controller goes to Settings > Security in Dashpivot, sets MCP Access to anything other than None and saves. The app won't authenticate until this is done.
-
Step 2
Add the Sitemate App in ChatGPT
In ChatGPT, go to Settings > Apps and Connections > Explore Apps. Search for Sitemate, open the result and click Connect. Sign in with your Sitemate credentials and approve the permissions.
-
Step 3
Check the Tools Are There
Open a new chat and look under the tools icon for the Sitemate tools. Sitemate MCP is read-only for now, so ChatGPT can't create, edit or delete anything in Dashpivot.
If the App Won't Appear
Three causes cover nearly every failed connection, and the first is the most common.
- Nothing connects: MCP Access is still off in Dashpivot, and only an Org Controller can turn it on
- Sitemate isn't in the list: you need ChatGPT Plus, Team or Enterprise, as connectors aren't on the free plan
- The tools don't show: open a new chat and check the tools icon
Where the SWMS and Its Review Live
Once it's reviewed, the SWMS still needs the crew's sign-on and somewhere to live. The review is worth more when it's kept with the record it checked.
-
Storm Captures It on Site
Fix the gaps, then save the SWMS as a PDF or Word file. Storm, Sitemate's data capture agent, rebuilds it as a Dashpivot template with AI Templates, and the crew signs on from a phone or tablet on site.
-
Dashpivot Holds It as the Record
Dashpivot stores each SWMS and its sign-on against the job as a structured, versioned record that's ready for audit. The version you reviewed is the version the crew signed.
-
Sitemate MCP Reads It Back
Each SWMS sits in Dashpivot, so ChatGPT can read across all of them over Sitemate MCP and answer questions like which jobs have a signed SWMS, or which gaps keep coming back. Access is read-only, and your existing permissions still apply.
Storm Fills the Records ChatGPT Reads
Sign-on happens on site, on a phone or tablet, before anyone starts the work. That puts the signed SWMS in Dashpivot the same day, and it's the same version ChatGPT reads back when you ask about the job.
AI Form Fill
Capture what you see and say, and Storm drafts the form for you. The details land in the right fields, so there is a record to check the moment the work is done, and you review and submit it yourself.
Listens to the work
Captures voice cleanly through machinery, wind, and movement. Accents and trade language understood. No headset needed.
Works on site
Built for gloves, cabs, and screen protectors. Captures inputs offline and processes in the background. No waiting on the field.
You stay in control
Storm never commits a record on its own. Every output is a draft for human review. Admins control what Storm can write to.
Frequently Asked Questions
Can ChatGPT sign off a SWMS?
No. It lists the gaps it finds. Deciding whether controls are adequate and signing off is the competent person’s job.
Does ChatGPT check the SWMS against the site?
No. It checks the document against the standard you give it. It doesn’t know the ground conditions or which other trades are on site, so you’ll still need your own site check.
Will ChatGPT catch every gap in a SWMS?
No. Use it as a quick first pass for common gaps, like a hazard without a control or a missing risk rating. Your own review still needs to happen.
Can ChatGPT write the SWMS as well as review it?
Yes, using separate drafting prompts. Describe the high-risk work, the site and the crew, and ChatGPT will draft it. A competent person still reviews and signs it before work starts.
Where does the reviewed SWMS end up?
On site, the crew signs onto it with Storm, Sitemate’s data capture agent, and Dashpivot by Sitemate stores the record. ChatGPT can read it back over Sitemate MCP, which is read-only, so it can’t change the record.