Damp Survey Report: Write Findings and Quote Lines With AI

Record your readings and observations in a fixed notes template, then have ChatGPT or Claude turn each finding into plain-English report text and a matching quote line from your rate card. The cause and the recommended works stay your own entries. The AI only describes what you recorded, and you check every line against your readings and photos.

Damp is a diagnosis problem before it is a writing problem. One house often has three causes at once, such as a bridged damp-proof course, a leaking gutter and a bathroom with no extraction, and the trade body's guidance describes investigation as a process of elimination. The AI should never be the one eliminating. What it can do is turn your conclusions into consistent wording, keep the report and the quote lined up item by item, and point out where your own notes contradict each other.

Follow me on Instagram@sagnikteaches

Notes first: a template the AI cannot misread

Most poor AI-written reports start with poor input: a page of shorthand such as "LR frt wall RD 28/22/18, salts +, GL high", with the conclusion buried in a sentence. The assistant will turn that into fluent paragraphs, and fluent paragraphs hide whatever you failed to record. Fix the input first. Use one template for every visit, filled in on site by typing or dictating into your phone, so each finding arrives with its readings, observations, conclusion and recommended works as separate fields.

Connect on LinkedInSagnik Bhattacharya

PROPERTY: [house type, age, wall construction]  VISIT DATE: [date]
WEATHER: [dry or wet in the last 48 hours]
INSTRUMENTS: [pin meter and its scale, hygrometer, thermal imager]

FINDING [number]: [room and wall, ceiling or floor]
Readings: [instrument | scale | height or position | value]  (one per line)
Observations: [what you saw, inside and outside, in plain words]
Photos: [numbers only]
MY CONCLUSION (cause): [rising damp | penetrating damp | condensation | other | not determined]
Confidence: [high | medium | low] because: [one line]
Recommended works: [codes from my works list, in order]
Quantities measured: [lengths, areas, numbers]
Not inspected or access limits: [what you could not see]

Two fields do most of the work. MY CONCLUSION and Confidence belong to you, and the drafting prompt tells the assistant to copy them word for word. The Photos field carries numbers only, because you write the captions yourself; stains photograph alike whatever caused them. If you dictate, read the figures back before you leave site, since dictation software can mangle numbers.

Subscribe on YouTube@sagnikteaches

Put the template, your works list (the standard repairs you sell, each with a code) and your rate card (code, unit, rate) into a Project in ChatGPT or Claude. Both let you keep files and standing instructions in a Project. A personal paid plan is $20 a month on either, and Claude's Free plan includes Projects too, up to five. Replace the homeowner's name, address and phone number with a job number before anything is uploaded.

One house, three causes: the readings behind the worked example

For the main example, take a two-storey house with solid brick walls and a rear kitchen extension. The homeowner reports a musty smell in the lounge, black mould in the first-floor bathroom and a stain in the kitchen corner. Three complaints, three different causes. The values below are illustrative figures on a pin meter's wood moisture equivalent (WME) scale; your own meter's scale and your firm's trigger values decide what counts as high.

Look at the patterns. In the lounge, readings fall with height, the shape surveyors associate with moisture drawn up from the ground. In the kitchen they do the opposite and peak near the eaves. In the bathroom the walls are largely dry and the story is in the hygrometer: at 17.5 °C and 78% relative humidity the dew point is about 13.6 °C, so a ceiling corner at 12.0 °C is below it and water will condense there. Some hygrometers display dew point directly; otherwise a psychrometric chart gives it.

From readings to findings: the prompt and the draft it returns

Give the assistant the filled-in template and a prompt that fixes what it may and may not do:

Using only the notes below, write the Findings section of a damp survey report for
the homeowner. For each finding: a heading with room and surface; what was measured
and where (units and instrument scale included); what was seen; the cause exactly as I
state it under MY CONCLUSION with the confidence I state; and my recommended works in
my order.

Rules
- Copy my conclusion and confidence. Do not add causes, readings or works.
- Never use the words cured, guaranteed, eradicated, permanent or safe.
- If two notes contradict each other, list them under QUERIES and do not resolve them.
- Plain English. Explain DPC and salts in a clause the first time.

[paste notes]

For finding 1, a draft in this spirit reads as follows:

Finding 1: Lounge, front wall. Readings taken with a pin-type meter on its wood moisture equivalent (WME) scale were 28% at 150 mm above the floor, 22% at 500 mm, 17% at 1,000 mm and 14% at 1,500 mm, so the wall is wettest at the bottom and drier towards the top. A salt test at skirting height was positive. Outside, the render on this elevation runs down past the line of the damp-proof course (the layer built into the wall to stop ground moisture rising) and the ground sits 60 mm above it. In our opinion this is rising damp with a bridged damp-proof course. Confidence: high.

It is accurate to the notes, which is the standard to hold it to. Three things still need fixing in a typical first pass:

• Confidence drifts into certainty. A first draft may write "confirmed" where your field says "high". Give the assistant a fixed vocabulary (high, medium and low, each with one stock phrase) and refuse anything else.
• Promises appear. Phrases such as "this will stop the damp returning" arrive from the assistant's general habits, not from your notes. The banned-word rule catches the obvious ones; the scan in a later section catches the rest.
• Contradictions are smoothed over. If finding 2 says "no ventilation" and also "trickle vent closed", a vent exists. The QUERIES rule makes the assistant ask you, instead of picking a story, and the answer is "no extract fan; trickle vent closed".

Cause and evidence: the one sentence the AI must not write

The cause line is the surveyor's professional judgement, so the assistant copies it and never improves it. Where AI does earn its place is as a second reader that challenges your conclusion. Use a separate prompt for that, in a fresh chat, so it does not simply agree with the report it has just written:

Here are my notes for finding 3, including my conclusion. List every reading or
observation that does not fit my conclusion, and every alternative cause my notes
cannot rule out. Also list any wall with internal damp but no external observation.
Do not tell me what the cause is.

A useful answer lists things like these: the notes say the gutter joint was leaking but not whether that was seen in rain or inferred from staining; there is no reading taken outside near the base of the wall, where the blocked downpipe shoe could be soaking the brickwork; and nothing records the condition of the lintel above the window. None of that changes your conclusion, but each item is a question you can answer before the report goes out, or an entry for the limitations paragraph.

The last rule in that prompt comes straight from the trade body's guidance, which says an external inspection should always form part of the diagnosis, working from the roof downwards, and that not every defect is visible without deeper investigation. Read the trade body's guidance on investigating dampness if you want your template's fields to follow it. The trade body's separate guidance on condensation notes that condensation inside the structure can easily be mistaken for rising or penetrating damp, which is one reason the AI must not rank causes from photographs.

If you use a thermal imager, its images belong in the notes as numbered photos with your own captions; turning thermal scans into plain-English findings covers that side. For a visit where mould is the main complaint, a mould inspection report template with findings, photos and limits has the sections to borrow.

Recommended works in the order they have to happen

Order matters to the result and to the quote. Stop the water first, treat the wall second, reinstate last. Plaster laid over a wall still being fed by a leaking gutter is money wasted, and a customer who does the work in the wrong order blames the firm. For the example house the order is:

1. Rear gutter joint and downpipe shoe (finding 3), because it stops water entering the wall.
2. Lower the ground and cut back the bridging render on the front elevation (finding 1).
3. Ducted humidistat extractor fan in the bathroom (finding 2).
4. Chemical DPC injection to the lounge front wall (finding 1), offered as an option.
5. Strip the lounge plaster and replaster with a specialist plaster system once the wall has dried (finding 1).
6. Patch replaster the kitchen corner after re-reading the wall (finding 3).

Item 4 is marked as an option for a reason. The trade body's guidance on rising damp says that where the damp-proof course is compromised by high ground levels, bridging render or debris, removing the bridge should hopefully be enough to control capillary rise. That is a call about when to inject and when to wait, and it is yours. The same guidance lists a chemical DPC as the usual economical treatment, says dry lining is not recommended as the primary or long-term remedy for rising damp in masonry, and expects affected plaster to be replaced with a specialist plaster. Put those preferences into the works list, then tell the assistant to recommend nothing that is not on it.

A short order check is worth running on every draft: "Check that no item that reinstates (plaster, paint) comes before the item that stops the cause. List any that do." Assistants tend to be reliable at this kind of sequence check because the answer is in the text you gave them.

From recommended works to quote lines the customer can follow

Each work item becomes a quote line with a code, a quantity, a unit and a rate. The assistant fills in the code and quantity from your notes. The rate comes from your rate card, and the arithmetic happens in your spreadsheet or quoting software, not in the chat. All rates here are placeholder list prices in USD.

Add it up yourself: finding 1 comes to $1,203.80, finding 2 to $346.00 and finding 3 to $466.00, so the quote totals $2,015.80. Because every line carries its finding number, the homeowner can accept the extractor fan and the gutter repair first and decide on the injection later. The two lengths on finding 1 are deliberate: the front elevation measures 5.4 m outside, while the lounge wall measures 5.0 m inside, and 5.0 m by 1.2 m gives the 6.0 m² of plaster. Record both in the Quantities field so the assistant never has to guess which one a line uses.

The draft the assistant returns will usually need correcting in these places:

• Wrong unit. The DPC injection came back in m² instead of linear metres. Injection is priced along the wall, so the unit check belongs in your rate card, where each code has one allowed unit.
• Wrong height. Replastering was priced to the full 2.4 m wall height (5.0 m by 2.4 m is 12.0 m², or $744) when your notes said 1.2 m (6.0 m², or $372). That is a $372 overstatement the customer would spot.
• A missing line. No skip. Hacking off plaster produces debris, and unpriced debris comes out of your margin.
• A conditional shown as certain. The injection is the surveyor's option, not part of the core price. The prompt should require "(option)" on any item you flag as one.

When the replastering lines need a closer look, a plastering quote template with skim, render and patch lines goes deeper on the unit rates.

Phrases that turn a damp report into a complaint

Customers keep damp reports for years and quote them back when a stain reappears. A few phrases create liabilities the survey never earned, and AI drafts reach for them because reassuring language is common in written reports. Keep a banned list in the Project and run a claims scan before sending: "List every sentence that promises an outcome, or states something we did not measure or inspect."

The safer column has something in common: each version states what you measured or did, and who should answer anything beyond it. That is easier to defend than reassurance, and it is also what makes the report credible to a homeowner who has read three conflicting opinions already.

The limitations paragraph: what one visit cannot show

A limitations paragraph written once and pasted into every report protects nobody. The paragraph has to describe the visit you made. Feed the assistant the Not inspected field from every finding and ask it to compile them, with the visit conditions from the top of the template. A good result reads like this:

Fitted wardrobes in bedroom 2 prevented access to the north wall, and floor coverings were not lifted. Readings were taken on one day after two dry days and show conditions on that day only. The surveyor did not open up the structure. Salts in plaster and metal items buried in walls, such as pipes and nails, can raise meter readings, so readings were read alongside what was seen on site and not on their own.

The last sentence is not boilerplate. One meter maker's own manual says readings are not definitive and only help a professional reach an informed judgement, that conductive material such as salts, carbon and metal can give false positive readings, and that its non-invasive search mode suits rapid surveys of solid walls, with deep-wall probes used to investigate any high readings it finds. The same maker's technical note adds that knowing the moisture percentage of a building material other than wood does not tell you whether it is wet or dry, which is why your report should cite the scale and your trigger values instead of asserting that any number is "damp".

Client data, the RICS standard and telling the homeowner

Three habits keep the process safe. First, send only what the assistant needs: a job number in place of a name and address, and photos with no identifying detail visible in frame. Consumer plans have a model-training switch in privacy settings, while business plans such as ChatGPT Business and Claude Team do not train on business content by default (both need at least two seats). Client-data basics for building surveyors covers the wider picture.

Second, if anyone in the firm is an RICS member or the firm is RICS-regulated, the RICS responsible-use-of-AI standard has applied since 9 March 2026. It covers AI output that has a material impact on a service, and its own examples include summarising documents that are then relied on in a report and composing significant parts of an opinion. Members must record a written decision on how reliable such output is, and tell clients in writing, in advance, when and for what purpose AI will be used. The register, assessment and client notice for small firms turns that into paperwork.

Third, whoever you are, tell customers in your terms. One sentence does it: "We use AI software to help draft report wording from the surveyor's own notes and readings. The surveyor checks and signs every report." Check with your professional indemnity insurer how it views AI-drafted reports, and keep the answer on file.

Checking the report before it goes out

Use the same six checks on every report, and tick them off in the Project's checklist:

1. Every reading in the report appears in your notes with the same value, unit and height.
2. Every conclusion and confidence word matches your fields exactly.
3. No banned word or promise survives the claims scan.
4. The works run in the right order, and no quote line lacks a finding number.
5. Quote quantities match your measured quantities, and the total was added in a spreadsheet or quoting tool.
6. The limitations paragraph describes this visit, not a generic one.

To see whether the routine pays back, time the next five reports the old way and the new way, and keep two counts: findings changed after the AI draft, and customer questions after sending. If one finding in three keeps changing, the fault is probably in the template or the works list, not the wording. Fix the input and the draft improves without touching the prompt.

Questions from damp and timber firms

Can I upload site photos for the AI to diagnose?

Not for diagnosis. Staining, mould and salts look alike whatever caused them, and a chatbot sees no readings, no gutter and no ground level. Use photos as evidence you have numbered and captioned yourself, and let the assistant place them in the report by number. If you ask it to describe a photo, treat the description as a draft caption and correct it against what you saw.

What if my site notes are handwritten?

Photograph the page and ask the assistant to transcribe it into the template, then check every number against the original. Handwriting recognition is usually more reliable with words than with digits and units, and a misread 17 for 11 changes a finding. Typed or dictated readings are safer, and reading the figures back before you leave the property takes under a minute.

Should the report and the quote be one document?

Many firms keep them separate and number the quote lines against the report's findings. A homeowner can then accept the gutter repair without committing to the injection, and a disputed finding does not hold up the whole job. Give the assistant the finding numbers as the key, so every quote line points back to exactly one finding.

How long should the homeowner summary be?

One page, around 250 to 300 words: what you found in each room, what caused it, what you recommend and in what order, and what the work does not cover. Explain terms such as damp-proof course in a clause the first time. Have the assistant write it last, from the finished findings, so it cannot add anything new.

Further reads

• Home Inspection Report Template: Build a Comment Library With AI — Build a comment library so recurring findings read the same each time.
• Can ChatGPT Write Home Inspection Report Narratives? What to Check — What to check when AI writes report narratives people rely on.
• Can AI Turn Drain Camera Notes Into Customer Reports? — Another notes-to-report workflow where a defect list drives the wording.
• How Surveyors Use AI to Turn Site Notes Into Reports — The general routine for turning site notes into a report.
• Snagging Report Template: Sections, Severity Order and AI Captions — Severity order and photo captions for defect-style reports.
• AI for Building Surveying: A Worked Fee Quote and Enquiry Reply — Quote your survey fee and reply to the enquiry with AI.
• Garage Conversion Leads: Qualify by Door Size, Floor and Planning With AI — A seven-question intake, a six-photo checklist and green, amber, red sorting that tells a garage conversion firm which enquiries deserve a survey visit.
• Home Energy Audit Report: Turn Findings Into a Client Report With AI — Turn audit notes into a client report with AI: a findings table, priority order, plain-English summary and savings as ranges, with every figure checked.
• Insulation Proposals: Explain Options With AI Without Promising Savings — Option-by-option wording for insulation proposals, with a plain-English R-value explanation and a claim check that keeps savings promises out.
• Insulation Quote Template: Material, Thickness and Area Lines With AI — A line-item insulation quote with thickness, area and exclusion wording, plus a prompt that flags missing measurements instead of inventing them.
• Sewer Camera Inspection Reports: From Defect List to Customer Summary — How a drain survey firm turns coded camera findings into a customer summary with AI, with a prompt, a worked defect list and the surveyor's checks.
• Solar Site Survey Checklist: Photos and Notes Sorted Into a Report With AI — Grouped photo checklist with why and how to verify each item, then a tagging system and prompts that sort photos and notes into a survey report.
• Structural Engineer Report Template: A Plain-English Summary You Can Check — A blank report skeleton, a client summary box, and a sentence-by-sentence check that stops AI changing the technical meaning, shown on a sample extract.
• Structural Engineer Inspection Report: Photos and Notes to a Draft With AI — A photo-to-report workflow for structural engineers: numbered photos, a register, AI descriptions of what is visible, and checks before the engineer signs.
• Loss Adjuster Software: Claim Files and Reports, and Where AI Fits — What a small adjusting firm needs from claims software, with verified prices and a clear line on where AI drafts and the adjuster signs.
• AI Due Diligence on Real Estate: Read Leases, Surveys, Searches — Upload the lease, survey and searches, build an issue list, rate the red flags and arrive at the solicitor with sharper questions. It never replaces advice.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: a moisture meter maker's user manual and technical note on WME readings; trade body guidance on investigating dampness, rising damp and condensation; RICS responsible-use-of-AI standard and announcement; ChatGPT and Claude plan pages.