Upload 200 photos. Categorised in 3 minutes.

A vision AI processes each photo and returns a RICS section code (D, E, F, G or I), element code (e.g. D2, F1), condition rating (1/2/3), professional caption, defect list with urgency tags and a survey-relevance flag. A scoring algorithm then picks the best 35 for your report, with guaranteed section coverage and a defect-priority boost for rating-3 elements.

Typically 2–3 minutes for a 200-photo upload. The vision AI processes images in parallel so turnaround does not increase linearly with photo count up to the 200-photo practical limit.

Every assignment is overridable. Drag photos between sections, mark front elevation, bulk-reassign with section dropdowns and toggle inclusion via the eye icon. The photo AI suggests; the surveyor signs off. Overrides are preserved through re-runs.

The practical limit is 200 photos per report, categorised in ~2–3 minutes. The selection step targets the best 35 images (configurable between 25 and 50 via the slider) regardless of upload size.

The defect list is generated from what the vision AI observes in each individual image. Vision AI tends to under-report rather than over-report defects, so you should still review the list — especially for photos where a condition rating of 3 (Serious Defects) was returned.

Yes. HEIC files are automatically converted to JPEG server-side before categorisation. No need to change camera settings on your phone before the inspection.

Each photo receives a survey-relevance flag. Furniture-only shots, ceiling or floor accidentals and unrelated images are excluded from the report selection by default. You can override any exclusion — useful if an apparent furniture shot is actually a damp-stain or defect photo.

Photos are scoped to a single report. For the same property you can access previously uploaded images via the property data drawer; the categorisation re-runs against the new report context rather than carrying over old assignments.

Yes. The photo categorisation pipeline runs identically for Level 3 Building Surveys. Element codes, condition ratings and defect lists flow into the expanded Level 3 report sections. The section-coverage guarantee and defect-priority boost apply equally — particularly useful for complex or older properties where Serious Defect (rating 3) elements are more common.

The vision AI identifies the dominant subject in each photo. If the room shot contains a clearly identifiable element — boiler, consumer unit, window, staircase — it routes to that element code. If the shot is genuinely room-level with no dominant subject, the AI routes to the most prominent element visible (e.g. E3 Internal walls for a general room shot). Confidence scores tend to be lower for these, surfacing them in the review queue. Drag-and-drop reassignment takes a few seconds if you prefer a different element.

No. Upload your photos in any order — straight from your phone or camera card. The vision AI determines the correct RICS element, condition rating and caption for each image automatically. The only preparation needed is to ensure HEIC files are not blocked by your device (they are converted automatically server-side).

Image AI supports RICS photo evidence requirements in two concrete ways. First, the missing-evidence checker alerts you if you have rated an element 2 or 3 but have not attached a supporting photo — so rated-but-unphotographed defects are caught before you sign off. Second, the section-coverage guarantee means every populated element section in the report has at least one photo, which avoids gaps in photographic evidence that could be questioned in a PI claim.

For a sole practitioner with no admin support, the biggest gain is eliminating the 80–120 minutes of manual photo sorting that otherwise happen between site and submission. Upload the batch, review the two or three low-confidence assignments, and the photos are ready to flow into the report writer — all before you have written the first element narrative. At typical survey volumes of two to three jobs per week, that recovery adds up to a working day each month.

Three features work together. The missing-evidence checker flags any element rated 2 or 3 without a supporting photo before you sign off the report. The section-coverage guarantee ensures every populated element section has at least one photo in the report appendix — no gaps that could be questioned in a dispute. The defect-priority boost automatically allocates at least two photos to sections with rating-3 (Serious Defects) elements, so your most significant findings carry the strongest visual evidence. RICS Home Survey Standard requires photographic evidence for rated defects; Image AI makes compliance automatic rather than a manual checklist item.

General-purpose photo AI assigns consumer labels — kitchen, garden, roof — which bear no relation to the RICS element catalogue. Image AI maps every photo against RICS element codes D1–D9, E1–E9, F1–F7 and G1–G3 specifically. It assigns a RICS condition rating (1/2/3), generates a survey-appropriate professional caption and extracts a defect list in surveyor language. The missing-evidence checker, section-coverage guarantee and defect-priority boost are also unique to the survey workflow — they exist because RICS photo evidence requirements are more precise than a generic image labelling task.

Image AI covers all 32 residential RICS element codes across five survey sections. Section D (external fabric): D1 Chimney stacks, D2 Roof coverings, D3 Rainwater fittings, D4 Main walls, D5 Windows, D6 External doors, D7 Conservatory/porches, D8 Other joinery and finishes, D9 Other external. Section E (internal condition): E1 Roof structure, E2 Ceilings, E3 Walls and partitions, E4 Floors, E5 Fireplaces, E6 Built-in fittings, E7 Woodwork, E8 Bathroom fittings, E9 Other internal. Section F (building services): F1 Electricity, F2 Gas/oil supply, F3 Water supply, F4 Heating, F5 Water heating, F6 Drainage, F7 Common services. Section G (grounds): G1 Garage, G2 Outbuildings, G3 Grounds and boundary. Section I (risks): I1 Risks to the building, I2 Risks to the grounds, I3 Risks to people, I4 Other risks. Photos of risk indicators — damp staining, structural cracks, visible asbestos-containing materials, flooded grounds — are routed to the I-section and flagged for the surveyor to verify.

Every photo receives a confidence score. Low-light images, shots at extreme angles (looking up at eaves, across a roof slope), and close-ups of narrow details all tend to produce lower confidence scores — typically below 70 — which surfaces them at the top of the review queue automatically. The AI still assigns the most likely element code but flags the assignment for the surveyor to confirm or override. For very dark or heavily blurred images where no building element can be identified, the photo is marked for manual assignment rather than guessed. The result: high-confidence assignments proceed without review, and genuinely ambiguous shots are the only ones that need your attention.

Image AI accepts JPEG, PNG, TIFF and HEIC. HEIC files from iPhone and iPad cameras are automatically converted to JPEG before categorisation — no need to change camera settings before the inspection. Standard Android camera JPEG and PNG files upload without conversion. RAW formats (CR2, NEF, ARW) are not currently supported; export to JPEG or TIFF from your desktop software before uploading.

Yes. Image AI categorises, captions and extracts defects from snagging inspection photos in the same way as RICS Level 2 and Level 3 photos — giving you a photo-by-photo snag record with per-defect urgency tags (immediate, short-term or routine). The missing-evidence checker also applies, so any element you have rated without a supporting photo is flagged before sign-off. For structuring and presenting the full snagging list to a developer or client — grouped by room, trade and priority — pair Image AI with the snagging report tool.

Yes. Image AI accepts JPEG, PNG and TIFF from any camera make or model — DSLR, mirrorless, compact or smartphone. HEIC files from iPhone and iPad convert automatically server-side; RAW formats (CR2, NEF, ARW) should be exported to JPEG or TIFF before uploading. A resolution of 2 MP or above is sufficient for reliable RICS element recognition. Very large files are reduced before processing so upload speed is not affected by shooting in maximum resolution.

Every assignment carries a confidence score from 0 to 100. Scores of 80 or above indicate a near-certain match — the element code, condition rating and caption are almost certainly correct and require no action. Scores between 50 and 79 are medium confidence, usually correct but worth a glance. Scores below 50 indicate genuine ambiguity — these surface at the top of the review queue so you can confirm, reassign or exclude the photo in seconds. In practice roughly 10 % of a 200-photo upload needs any manual attention; the rest flows through automatically.

Roughly 90% of a typical 200-photo upload receives a high-confidence assignment (score 80 or above), meaning the element code, condition rating and caption are almost certainly correct with no review needed. Around 8–10% score in the medium-confidence range (50–79) and are worth a glance. Fewer than 5% score below 50 and surface in the review queue as genuine ambiguity cases. The AI is trained specifically on RICS survey contexts — chimney stacks, roof coverings, consumer units, damp patches, bathroom fittings — rather than general photography subjects, which is why accuracy is high for standard survey subjects but tends to dip for unusual or rare elements such as historic materials or very tight-access spaces. The confidence score itself is the best guide: high-confidence assignments are reliable; the queue shows you exactly which ones to verify.

Yes. The AI identifies each photo by visual subject — not by construction method or material — so RICS element codes apply regardless of building age or materials. A thatched roof is still D2 Roof coverings; a stone chimney stack is still D1 Chimney stacks; exposed timber-frame infill is still D4 Main walls. For unusual or rare subjects where the element is harder to identify clearly, the confidence score tends to be lower, which surfaces those photos in the review queue for you to confirm or reassign. Drag-and-drop reassignment resolves most cases in seconds. The full D1–D9, E1–E9, F1–F7 and G1–G3 catalogue applies to properties of any age or construction type.

Yes. For flats and maisonettes the pipeline adapts automatically. Elements that are not present or not inspectable — chimney stacks, external roof slope, grounds — simply produce no photos; the AI does not flag their absence as a problem. Common flat elements — F1 Electricity, F4 Heating, E-section internal condition, F6 Drainage — categorise identically to a house survey. The missing-evidence checker only flags elements you have rated in the report, so uninspected or inaccessible elements are not incorrectly marked as photographic gaps. This makes Image AI equally useful for flat surveys, maisonettes, apartments and upper-floor conversions.

The vision AI identifies damp staining as a visible defect regardless of which element the photo shows. A damp patch on an internal wall routes to E3 Walls and partitions with "damp staining" in the defect list and an urgency tag of immediate or short-term depending on severity. Penetrating damp on an external wall or render routes to D4 Main walls. Ceiling staining from a roof or bathroom leak routes to E2 Ceilings. Photos where damp is the dominant subject tend to receive high confidence scores because the defect is visually clear — and a condition rating of 2 or 3 depending on the extent and spread visible in the image. The missing-evidence checker will also flag any element where your written findings mention damp but no supporting photo has been attached.

Yes — they run at different stages of the same report. Image AI runs first, categorising and selecting photos before or during element write-ups. Once your element narratives are drafted, the RICS Standards Checker reviews the full report text for rating contradictions, missing specialist referrals and language that falls outside the RICS Home Survey Standard. Running Image AI first means your photo evidence and written findings are already aligned before the compliance pass — reducing the chance of a photo-rated element conflicting with your written condition rating and flagging a contradiction that requires correction.

Photos upload directly to secure cloud storage linked to your Home.co.uk surveyor account. Access is restricted to your own account and the other surveyors in your organisation — your client property photos are never visible to other practices or third parties. Each photo is scoped to the specific report it was uploaded to and is not shared or indexed externally. If you need to remove photos after a report is completed — for data retention or client request purposes — the Photos tab lets you delete individual images or the full batch, which removes both the stored photo and its AI categorisation record from your account.

Yes. Every report you produce with Image AI creates a structured photo evidence trail — RICS element codes, condition ratings, captions and defect lists — that documents how photographic evidence was gathered and linked to your written findings. For APC candidates working toward Building Surveying or Residential Survey competencies, the missing-evidence checker demonstrates a quality-control step applied before sign-off: rated-but-unphotographed elements are flagged and must be resolved, showing professional diligence rather than accidental omission. The section-coverage guarantee records that photographic evidence was applied systematically across all populated elements rather than selectively. The override history — photos the AI assigned one way and you reassigned — shows AI assistance was used and critically reviewed, not relied upon uncritically. This documented approach supports the competency strand of demonstrating professional judgement in AI-assisted work that assessors increasingly expect to see addressed.