AI property capture case study

Building Snapsure: from photo to compliant record, in one capture.

How Changeable designed and built Snapsure, an AI-powered capture-to-compliant-record platform that turns photos and voice notes from a property inspection into structured, legally formatted documentation, without a single form.

ProductSnapsure Capture-to-Compliance Platform
ApproachOne capture engine, expanded across use cases

Project overview

Snap it. Speak it. Sorted.

Property inspections in New Zealand still largely run on paper forms or clunky templated apps: a person walks through a property, manually selects a condition rating from a dropdown for every single item, and types notes by hand. Changeable built Snapsure to remove that entire layer of friction. A user photographs a room and narrates what they see; AI classifies the condition of every item, structures the narration into the right fields, and produces a compliant, professional report ready for signature.

The build deliberately started narrow, a single New Zealand tenancy inspection workflow, before expanding the same underlying capture engine into compliance assessment, real estate listing content and multi-user agency use.

Capture-first, not form-first

No dropdowns to work through. A photo and a sentence replace a full paper checklist.

One engine, several outputs

The same photo and voice capture powers inspections, compliance checks and property listings.

Compliance-aware by design

Report formats and required disclosures are built against actual New Zealand tenancy legislation.

Tested against a structured story set

The build was validated against 58 defined user stories covering the full inspection lifecycle before release.

The problem

Existing tools bolt AI onto a form. Nobody had built AI-native from the start.

New Zealand already has property inspection apps. Most are built around a traditional form-based workflow, walk into a room, select each item from a list, choose a condition rating, type a note, move to the next item, with AI added afterwards as an assist feature rather than the core interaction.

That leaves the actual bottleneck untouched: the time spent manually working through structured fields room by room. The opportunity was a genuinely capture-first product, where the form disappears entirely behind a photo and a sentence.

What the market gap looked like

  • Established NZ tools remain form-based, with AI notes added as a bolt-on feature
  • Landlords and property managers still spend real time on manual data entry per item
  • Healthy Homes and other compliance requirements are often handled as a separate process
  • No existing tool used the same capture data for both inspection and listing content
  • Multi-property landlords and agencies had no single tool spanning solo use through to team use

What Snapsure does today

One capture engine, four connected capabilities

Every capability shares the same underlying photo and voice capture, extended with a different AI output layer for each use case.

Guided inspection capture

Room-by-room walkthrough with photo, video and voice narration, automatically tagged by room and timestamp.

AI condition classification

Photos are analysed for item type and condition, with damage such as stains, cracks and mould flagged automatically.

Compliance reporting

Reports formatted to New Zealand Residential Tenancies Act requirements, with e-signature capture built in.

Healthy Homes assessment

A guided six-standard compliance check covering heating, insulation, ventilation, moisture, draught stopping and smoke alarms.

Inspection comparison

Routine and exit inspections are compared against the entry baseline, with AI highlighting genuine condition changes.

Listing mode

The same capture engine generates a portal-ready listing description, with AI photo scoring to select the best shots.

Team and agency tools

Shared properties, role-based access and white-labelled reports for property management teams.

CRM integration

Listings can be pushed directly to a connected real estate CRM once created.

How it was built

A single capture engine, built once and extended

Rather than build separate tools for inspections, compliance and listings, Snapsure was designed around one shared capture layer from the very first build session.

Phase 1 – Prove the core capture and classification loop

Photo in, structured condition data out

The foundation was the app shell, property setup and the room-by-room capture flow itself: photo upload, voice recording and an initial keyword-based parser for spoken condition language, paired with AI photo analysis to classify item type and condition. Getting this loop right, so a genuinely natural sentence like “the carpet’s a bit worn” reliably became a structured, correctly categorised record, was the single most important thing to prove before building anything else on top of it.

Phase 2 – Turn captures into a compliant, signable report

Review, generate, sign, finalise

Once capture and classification worked, the build moved to what a landlord or property manager actually needs at the end: a review screen showing every room and item, a professional PDF report formatted against New Zealand tenancy documentation standards, and an on-device signature flow for both landlord and tenant. This closed the loop from a five-minute walkthrough to a finished, legally usable document.

Phase 3 – Extend into dedicated compliance assessment

A guided Healthy Homes standard, not a generic checklist

New Zealand’s Healthy Homes Standards were built as their own guided six-step assessment, heating, insulation, ventilation, moisture and drainage, draught stopping, and smoke alarms, each with its own compliance logic and photo capture. A compliance scorecard summarises the result per standard, with a dedicated report format and the specific retention disclosure required under the Residential Tenancies Act.

Phase 4 – Reuse the same engine for real estate listings

One capture flow, a different AI output layer

Rather than build a second capture experience, listing mode reused the exact same room-by-room photo and voice flow, then applied a different AI layer on top: portal-aware description generation, AI-scored best-shot selection, and simple photo enhancement. This is the clearest expression of the platform’s core idea: the capture engine is the product, and each use case is a different output built on it.

Phase 5 – Add team, compliance and integration depth

Agency tier, white-labelling and CRM push

The final layer of the build added what agencies and property management teams specifically need: shared team access with role-based permissions, white-labelled reports carrying an agency’s own branding, CSV data export, and a direct integration to push completed listings into a connected CRM. Security review was folded into this same phase, adding authentication checks to every AI-powered function before agency data started flowing through the platform.

Testing discipline

58 defined user stories, tested before a single external user touched the app.

Before Snapsure moved toward early access, the full product was validated against a structured set of real usage scenarios rather than informal spot-checking, the same discipline Changeable applies to AI governance and quality assurance in client work.

58 user stories defined across the full lifecycle, from sign-in through capture to signed report
Every defect logged, severity-rated and sequenced into a prioritised fix backlog
Compliance-specific testing checked report content directly against Residential Tenancies Act requirements
Real-device mobile and offline-capture testing scheduled as a distinct gate before field use
A dedicated pre-release plan gated data ownership, hosting, authentication and legal pages before any external tester
Voice-based in-app feedback built so early testers could report issues without leaving the workflow

Results

What the finished platform delivers

The build produced a genuinely dual-mode platform, not a single-purpose inspection tool with extra features bolted on.

Minutes, not a form

A full room inspection is captured through photo and speech, with AI handling the structuring that would otherwise be manual data entry.

One data set, two outcomes

The same property capture can produce a compliant inspection report or a marketing-ready listing, without recapturing anything.

Purpose-built compliance logic

Healthy Homes assessment reflects the actual six-standard structure landlords are legally required to meet.

Own infrastructure, not a demo tool

The platform runs on its own database and hosting, independent of the no-code tool used to prototype it.

Agency-ready from day one

Team roles, white-labelled reports and CRM push were built as first-class features, not a later add-on.

Structured path to release

A defined, gated pre-release plan sequenced security, legal and infrastructure readiness before any external user.

Questions

Questions about how Snapsure was built?

Common questions about the approach behind Snapsure’s capture-to-compliance platform.

Why build a capture engine first instead of the full feature set?

The riskiest and most valuable part of the product was proving that natural photos and speech could reliably become accurate, structured condition data. Every other capability, compliance reporting, listings, team tools, only makes sense once that core loop genuinely works.

How is this different from existing property inspection apps?

Most existing NZ tools are form-based, with AI added as an assist feature on top. Snapsure was designed capture-first from the start, so the form effectively disappears behind a photo and a sentence rather than AI simply speeding up form completion.

Does the same capture really work for both inspections and listings?

Yes. Listing mode reuses the identical room-by-room photo and voice capture flow used for inspections. The difference is entirely in the AI output layer applied afterwards, description generation and photo scoring for listings, condition classification and compliance formatting for inspections.

How were the Healthy Homes Standards built into the product?

As a dedicated six-step guided assessment covering heating, insulation, ventilation, moisture and drainage, draught stopping and smoke alarms, each with its own compliance logic, rather than treating Healthy Homes as a generic checklist bolted onto the standard inspection flow.

How was the platform tested before release?

Against a structured set of 58 defined user stories covering the full product lifecycle, with every defect logged, rated by severity and sequenced into a prioritised fix backlog before mobile and field testing began.

What happened before real external testers were invited in?

A dedicated pre-release plan moved the platform onto its own database and hosting, added real authentication and error tracking, and required privacy and terms documentation to be published and linked at sign-up before any invite went out.

Can Changeable build something similar for our organisation?

Yes. Snapsure reflects the same use case-led, governed approach Changeable applies in client engagements, including use case development, AI app development and AI governance.

Want to see how Snapsure turns a walkthrough into a report?

Explore Snapsure directly, or talk to Changeable about applying the same capture-first, governed approach to your own AI product or workflow.