Turning Your Answers Into a Snow Removal Process Document Using AI

THE PROBLEM:

Knowledge That Lives in Heads, Not Documents

You just worked through the five questions. Maybe you sat down with your team and talked through each one. Maybe you even typed out answers. You know how your snow removal operation should work.

But here's what happens next in most organizations: someone volunteers to "write it all up." Three weeks later, they've documented the vendor list and maybe the dispatch triggers. The role clarity section is half-finished. The site-specific requirements are "still being gathered." By the time snow actually hits, you're back to the same improvised chaos because the document never got finished.

Or worse, someone does finish it. They spend forty hours writing a comprehensive snow removal manual. It's thorough. It's detailed. And it's immediately outdated because they documented the process exactly as it existed in October, but by December, two vendors changed, you adjusted your trigger points after the first storm, and nobody has time to update the document.

The person who knows how your snow operation runs can't clone themselves. And they definitely can't spend a week every fall writing documentation that's obsolete before the first plow hits pavement.

THE PROMPT:

What to Feed the AI

Here's the prompt structure that takes your interview transcript or typed answers and generates a usable snow removal process document. You can copy this, adjust the bracketed sections for your specifics, and use it with any AI tool (Claude, ChatGPT, or whatever you prefer).

The Prompt:

I need you to create a comprehensive snow removal operations process document for a facilities maintenance department managing [NUMBER] convenience store/gas station locations across [REGIONS/STATES].

I'm going to provide you with either:

• A transcript of a conversation where our team discussed how we handle snow removal operations, OR

• Written answers to key questions about our snow removal process

Your job is to take that raw information and structure it into a clear, actionable process document that someone new to our operation could use to manage snow removal independently.

Required sections for the document:

1. Weather Monitoring and Pre-Storm Preparation

• How we monitor forecasts and what sources we use

• Trigger points for different actions (monitoring vs. pre-positioning vs. active dispatch)

• Pre-storm communication protocols (who gets notified, when, with what information)

• Vendor readiness verification steps

2. Dispatch Decision-Making and Execution

Specific trigger points for dispatch (accumulation thresholds, ice conditions, timing considerations)

• Who has authority to authorize dispatch and under what conditions

• Step-by-step dispatch protocol (what information to gather, how to contact vendors, what to log)

• How we handle simultaneous storms across multiple regions

3. Vendor Management During Active Storms

• Which vendors are assigned to which locations (create a clear mapping)

• Primary contact protocols and backup contacts

• Expected response times and arrival confirmation procedures

• What to do when vendors don't respond or can't meet commitments

• Emergency escalation process and backup vendor activation

4. Service Standards and Site-Specific Requirements

• Definition of "complete" service for our locations

• Priority areas that must be cleared (fuel islands, main entrances, pedestrian paths)

• No-plow zones and restricted areas (tank fill ports, utility access, specific landscaping)

• Salt/ice melt application standards

• Any location-specific variations or special requirements

5. Documentation and Quality Verification

• Required photos and timestamps for work order completion

• What information must be logged in our CMMS/work order system

• Quality verification steps (who checks, what they look for, what happens if service is incomplete)

• How we track vendor performance during storms

6. Post-Storm Process and Damage Documentation

• Post-storm inspection requirements

• How to document and report plow damage (bollards, curbs, signage, landscaping)

• Process for addressing incomplete service or quality issues

• Communication with store operations after storms

7. Role Assignments and Communication Protocols

• Who owns each phase of the storm response

• Decision-making authority at each level

• Communication flow during active storms (who updates whom, how often, through what channels)

• Escalation paths when problems arise

Formatting requirements:

• Use clear headers and subheaders for easy navigation

• Include specific decision points with "IF/THEN" logic where appropriate

• Call out any gaps where we need to add more specific information

• Use bullet points for lists and steps, but full paragraphs to explain reasoning or context

• Flag any areas where our current process seems unclear or incomplete

What to avoid:

• Don't make up details we didn't provide

• Don't use generic facilities management jargon—use our specific terminology

• Don't create overly complicated flowcharts or diagrams (stick to clear written process)

• Don't assume standard industry practices unless we specifically mentioned them

Input to process: [PASTE YOUR INTERVIEW TRANSCRIPT OR TYPED ANSWERS HERE]

Create a process document that's immediately usable but also identifies where we need to add more detail or make decisions we haven't clarified yet.

What to Include With the Prompt

Along with the prompt itself, you need to provide the raw material. This is either:

Option 1: Your Interview Transcript

If you recorded a conversation where you and your team talked through the five questions, get that audio transcribed. Most AI tools can transcribe directly, or you can use a service like Otter.ai or Rev. The transcript doesn't need to be perfectly clean. AI can work with conversational language, tangents, and even incomplete thoughts. Just make sure the core information from your five-question discussion is captured.

Option 2: Your Typed Answers

If you wrote out answers to the five questions, that works just as well. Don't worry about perfect formatting or complete sentences. The AI will structure it. Just make sure you've addressed:

• Who does what when weather hits

• What your dispatch triggers are and who authorizes dispatch

• Which vendors cover which locations

• What "complete" service means for your locations

• What gets documented and how

Optional Additions: If you have any existing documentation even outdated or incomplete, include it. Vendor lists, old contracts with service specifications, site maps, previous work order templates. The AI can incorporate that information into the new process document, which saves you from retyping details that haven't changed.

EXAMPLE:

What Good Output Looks Like

Let's look at what happens when you feed the prompt properly. Here's a before and after.

Before: Raw Interview Transcript Snippet

"So basically when we see snow coming, somebody needs to check the forecast, probably the night before. We usually look at Weather.com or sometimes the Weather Channel app. If it's gonna be more than like 2 inches we call the vendors. Well, sometimes 2 inches, sometimes less if it's gonna be during the morning rush. Really depends. For the Midwest stores that's Jim's Plowing and for the Northeast we've got three vendors I think? Wait, two now because we dropped that one guy last year. Anyway, we call them and make sure they're ready. Sometimes they don't answer right away which is annoying. We should probably have a backup for that..."

After: Structured Process Document Output

Weather Monitoring and Pre-Storm Preparation

The maintenance coordinator on duty monitors weather forecasts for all operating regions beginning 48 hours before any forecasted snow event. Primary forecast sources include Weather.com and The Weather Channel mobile app.

Monitoring trigger points:

• 48 hours before forecasted snow: Begin daily forecast checks

• 24 hours before forecasted snow: Verify vendor availability and contact information

• 12 hours before forecasted snow (if 2+ inches predicted): Initiate vendor pre-positioning communication

Vendor notification protocol: The coordinator contacts vendors once accumulation forecasts meet the following thresholds:

• 2+ inches during standard hours (10 AM - 8 PM)

• 1.5+ inches during morning rush (5 AM - 10 AM)

• Any ice accumulation regardless of snow depth

Current vendor assignments:

• Midwest region: Jim's Plowing (Primary contact: Jim Reynolds, 555-0123)

• Northeast region: [NEED TO CLARIFY - mentioned "two vendors now" but specific companies and contacts not provided]

Gap identified: No backup vendor contact process defined for when primary vendors don't respond. Current process: "Sometimes they don't answer right away which is annoying."

Action needed: Define maximum acceptable response time and identify backup vendors for each region.

See what happened?

The AI took conversational, incomplete information and:

1. Created logical structure - Organized scattered thoughts into clear sections with headers

2. Identified decision points - Pulled out the specific thresholds (2 inches, 1.5 inches, timing considerations)

3. Assigned ownership - Made clear who's responsible ("the maintenance coordinator on duty")

4. Flagged gaps - Called out where information was incomplete or unclear

5. Preserved specifics - Kept exact details like vendor names and threshold numbers

What the AI Did Well:

The output turned "somebody needs to check" into "the maintenance coordinator on duty monitors," which provides the specificity you need for process documentation. It took rough timing considerations ("probably the night before") and created a 48-hour/24-hour/12-hour monitoring cadence. It caught that you mentioned dropping a vendor but didn't provide the current Northeast vendor list, and flagged it as a gap to fill.

Common Issues to Watch For and Fix:

Even with a good prompt, AI output needs human refinement. Here's what to watch for:

AI being too generic.

If the output says "contact the appropriate vendor" instead of "call Jim Reynolds at Jim's Plowing (555-0123) and confirm availability for the forecasted storm window," make it specific. Generic process documents don't help anyone.

Missing your company-specific terminology or tools.

If you use ServiceChannel or Corrigo or a custom CMMS, make sure that's reflected. If you call your maintenance coordinators "facilities dispatchers" or "VMCs," make sure the AI uses your terminology. You can add a note to the prompt: "We use [CMMS NAME] for work orders and refer to our coordination team as [ROLE NAME]."

Over-complicating simple steps.

Sometimes AI will turn "call the vendor" into a five-step communication protocol with templates and follow-up schedules. If something is straightforward, simplify it. Your process document should be clear, not bureaucratic.

Assuming standard practices you don't actually follow.

If the AI adds a step you didn't mention like "conduct post-storm debrief meetings" and you don't actually do that, remove it. Document the process you'll actually execute, not the process that sounds impressive.

HOW TO ITERATE:

Feeding the Output Back With Refinements

Your first AI-generated draft won't be perfect. That's expected. Here's how to refine it:

1. Read through the entire document and highlight gaps. Where did the AI say "NEED TO CLARIFY" or "Action needed"? Those are places where your original answers were incomplete.

2. Fill in the gaps with specific information. Write out the missing details exact vendor names, complete contact lists, specific site requirements you forgot to mention.

3. Feed it back to the AI with refinement instructions. Copy the draft document, add your new information, and give the AI a refined prompt:

"I've reviewed the snow removal process document you created. I'm providing additional information to fill gaps and correct areas that need more specificity. Please update the document by incorporating this new information while maintaining the same structure and format:

[PASTE YOUR ADDITIONS AND CORRECTIONS]

Update the document and remove any 'gap identified' or 'action needed' notes where I've now provided the missing information."

You can iterate this way 2-3 times until you have a complete, accurate process document. Each iteration takes minutes, not days.

What You Have Now

At this point, you have a complete snow removal process document. Someone could read it and understand:

• When to start monitoring weather

• What triggers dispatch decisions

• Who to call and how to escalate

• What "complete" service means

• What needs documentation

That document is immediately more useful than the scattered knowledge that lived in your head last week. But a process document by itself doesn't execute. You still need to answer: who's doing what, specifically?