On InspectionX, the AI autonomously arranges your report as you inspect. When you submit a photo and a voice memo, the software scans your template and figures out which section, subsection, and comment best matches your observation. A well-structured template means the AI gets it right most of the time, which means faster report delivery and less editing.
When the template is not set up well, the software finds similar-looking sections, overlapping subsections, and comments that could apply to multiple findings. That is when misclassifications happen.
This guide covers the principles that mitigate those problems.
Table of contents
- How the AI categorizes your observations
- Every section should have a unique name
- Every subsection should exist in only one place
- Each section should only cover a single system
- Simplify the names of sections and subsections
- Each comment should cover one finding
- Have a comment for each severity level
- Quick checklist
How the AI categorizes your observations
When you submit a picture and a voice memo or note, the AI does three things in order:
- Finds the right section — which section from the template is most relevant to the audio or note? (Roofing, Kitchen, Electrical, etc.)
- Finds the right subsection — given that section, which subsection from the template is most relevant to the audio or note? (Gutters, Countertops, GFCI outlets, etc.)
- Selects the right comment — given all comments available for that subsection, which comment best matches what was described?
After it has done all three, the finding is published in your observation screen, fully classified and ready to go. The key is making sure your template gives the AI a clear, unambiguous path at every step.
Think of your template as a map. If two roads have the same name, the AI does not know which one to take. Make every path unique and the AI will get where it needs to go.
Every section should have a unique name
This is the simplest rule and the easiest one to overlook. If two sections in your template have the same name or a very similar name, the AI cannot tell them apart.
Let’s say you make the following observation: “the GFCI in the kitchen is defective.” The AI needs to find the Kitchen section. But if your template has both a section called Kitchen and another called Kitchen(s), the software does not know which one you mean.
Ambiguous sections
Kitchen and Kitchen(s) — which one should the AI choose?
Electrical and Electrical System — same problem
Clear sections
Each section has one unique name: Kitchen, Electrical, Roofing
No duplicates, no variations — one name per section
The fix: make each section name distinguishable from every other section. Merge duplicates, remove variations, and make sure no two sections could be confused for each other.
Every subsection should exist in only one place
This is the most common source of misclassifications. Each part of the house that gets inspected should only appear in a single place in your template.
Let’s say you are inspecting the gutters and you observe: “gutters are filled with debris.” If your template has a Gutters subsection under both Roofing and Exterior Elements, the AI does not know where to place it. Some reports it might go under Roofing, other reports under Exterior — and now you have inconsistency across your reports.
Duplicated subsection
Roofing → Gutters
Exterior Elements → Gutters
“Gutters are filled with debris” — the AI has two valid places to put this
Single location
Roofing → Gutters
Removed from Exterior Elements — now the AI only has one place to look
The fix: go through your template and look for any subsection that appears under more than one section. Pick the section where it makes the most sense and remove it from the other. Now the AI only has one place to place the finding in.
Each section should only cover a single system
When a section covers more than one system, the AI has a harder time matching your observations to the right place. It also makes your report harder for readers to navigate.
A common example is combining heating and cooling into one section. If you say “the AC is not cooling properly,” the AI has to sort through a section that also contains furnace content. Breaking them apart gives the AI a cleaner target.
Combined systems
Heating, Ventilation, and Air Conditioning → contains AC, furnace, ductwork, thermostat all in one section
Separated systems
AC — its own section
Furnace — its own section
Each system has a clear, dedicated place in the template
The same idea applies to interior areas. Instead of putting kitchens and bathrooms inside a single “Interior Elements” section, consider breaking them apart: one section for Interior Elements, one for Kitchen, and one for Bathroom. The more focused each section is, the more accurate the AI becomes.
Simplify the names of sections and subsections
Long, formal names make it harder for the AI to match your spoken observations to the right section. When you say “the AC is not cooling,” the AI has to figure out that you mean “Heating, Ventilation, and Air Conditioning.” A simpler name makes that match instant.
Overly formal names
Heating, Ventilation, and Air Conditioning
Electrical Distribution and Branch Circuits
Domestic Water Supply and Distribution
Simple names
AC
Furnace
Electrical
Plumbing
Use the words you would naturally say on-site
The rule of thumb: name your sections and subsections the way you would say them out loud during an inspection.
Each comment should cover one finding
Every comment in your library should describe a single finding. When a comment covers multiple issues at once, the AI cannot separate them from each other and it will bring in a comment into your observation that says more than you might want it to.
Let’s say you observe: “the sink is leaking in the kitchen.” If the matching comment in your template covers both a leaking sink and a clogged sink in the same entry, the AI is forced to select a comment that includes information about a clog — even though that is not what you observed.
Combined comment
“Sink is leaking and/or clogged. Recommend repair.”
If you only observed a leak, the report now mentions a clog too
Separated comments
“Sink is leaking. Recommend repair by a licensed plumber.”
“Sink is clogged. Recommend clearing the drain.”
If you need both, you can always create multiple observations
Our suggestion: one finding per comment. If two issues sometimes appear together, keep them as separate comments and create two observations when needed.
Have a comment for each severity level
For findings where severity matters, your template should have a separate comment for each severity level.
Think about stucco cracks. A hairline crack is cosmetic. A moderate crack might need monitoring. A significant crack with displacement could indicate structural movement. These are not the same finding — they carry different implications and need different language in the report.
One comment for all severities
“Cracks observed in stucco. Recommend repair.”
Does not distinguish between a hairline crack and possible structural movement
Separate comments by severity
“Hairline cracks observed in stucco. These are cosmetic and common with normal settling.”
“Moderate cracks observed in stucco. Recommend sealing to prevent moisture intrusion.”
“Significant cracks with displacement observed in stucco, suggesting possible structural movement. Recommend evaluation by a structural engineer.”
The same principle applies anywhere severity changes the meaning of the finding. Here is another example with moisture and stains:
Moisture comments by severity
“Dry stain observed on ceiling. No signs of active moisture. Appears inactive.”
“Stain with elevated moisture readings detected. Possible active leak. Recommend further investigation.”
“Active leak observed with visible fungi growth. Recommend immediate remediation by a qualified professional.”
When severity is clearly stated in each comment, the AI can match your observation to the right one based on what you describe on-site. If you say “hairline crack in stucco,” the AI selects the cosmetic comment. If you say “significant crack with displacement,” it selects the structural one. No guessing needed.
This also means that when you record your audio or write your note, you should mention the severity level. The more specific you are about the severity, the easier it is for the AI to pick the right comment from your library.
Wherever severity changes the meaning or the recommendation, create a separate comment for each level. This gives the AI — and your report readers — the right language every time.
Quick checklist
Take a few minutes to review your template. Look for these things:
- Duplicated sections — do any two sections have the same or very similar names? Merge them into one.
- Duplicated subsections — does any subsection appear under more than one section? Remove it from all but one.
- Multi-system sections — does any section cover more than one system? Break it into separate, focused sections.
- Overly formal names — are your section names longer than what you would naturally say on-site? Simplify them.
- Combined comments — does any comment cover more than one finding? Split it into separate comments, one per issue.
- Missing severity levels — for findings where severity matters, do you have a comment for each level? Add them where they are missing.
If your template follows these guidelines, InspectionX will pick the correct section, subsection, and comment every time. You will spend less time editing after the inspection and your clients will receive their report faster.
Your template is the foundation. The cleaner it is, the less work you do after the inspection — and the faster your reports get delivered.
Ready to build your template?
We can help you set up your template so the AI works for you from day one.
Schedule a Demo Contact Us