How Pristine Properties Captures Better Quotes — Before Anyone Gets On Site

When we last wrote about Pristine Properties, the story was about how ServiceM8 replaced a whiteboard and a pile of paper job cards. That was the foundation. What Basil Saab and the team have built on top of it since then is something more targeted: a quote capture tool that gathers structured requirements — from both inbound clients and the team's own staff — and uses AI to translate the details into every system Pristine Properties runs.

It doesn't replace on-site assessments. It doesn't remove the back office from the picture. What it does is make sure that by the time anyone picks up the phone or walks onto a site, the job is already half-set-up.

The Problem with Traditional Lead Capture

Before this system, the inbound enquiry process at Pristine Properties looked like most service businesses: a contact form, an email notification, someone checking the inbox, a phone call to scope the job, internal discussion about pricing, a manually written quote in Xero, and finally a follow-up email to the client. Start to finish, that could easily be a 24–48 hour round trip — longer if it landed on a Friday afternoon.

For a commercial property services business competing across Greater Sydney, that lag is a real cost. Clients requesting a commercial clean or a maintenance contract aren't waiting two days. They're getting three quotes, and whoever responds professionally and fast tends to win.

The other issue was consistency. Quote scoping depended on who took the call. Different staff interpreted requirements differently. Pricing wasn't always applied uniformly. The occasional detail — access requirements, floor area, specific cleaning products for a sensitive site — would get lost between the phone call and the written quote.

Three Steps. Simple Enough That the Team Uses It Too.

The solution was a purpose-built quote capture interface — accessible on the Pristine Properties website for inbound clients, and equally used by the team when they're scoping a job in the field or over the phone. It was designed to be simple enough that anyone could use it without training. In practice, that's exactly what happened: staff adopted it as their default quoting tool because it's faster and more consistent than any previous method.

From anyone's perspective — client or technician — it's three steps:

Step 1 — Select the Options, Then Describe the Details

The interface opens with a set of structured selectors — not an open-ended blank page. The user picks from defined options for the things that are always categorical:

• Property type — office, warehouse, strata, retail, residential, other
• Service type — regular clean, one-off deep clean, end-of-lease, maintenance visit, post-construction, other
• When — This week, next week, whenever.

These choices anchor the quote. They're consistent regardless of who's filling in the form — a client doing it themselves, or a technician capturing a job over the phone.

Below the selectors, there are free-text fields for the details that don't fit neatly into a dropdown: floor area, specific rooms or zones, access requirements, product preferences, timing constraints, anything unusual about the site. This is where the AI earns its place. The entries don't need to be formatted, precise, or structured — "big open warehouse, mostly epoxy floors, forklift access only through the side gate, they're fussy about the bathrooms" is enough for the system to work with.

Step 2 — AI Reads the Free Text and Structures It

Once the form is submitted, the structured selections and the free-text entries are passed together to an AI processing pipeline built with n8n, connected to our custom quoting code for ServceM8. The selections give the AI a confident starting point. The free text is where it does the interpretive work.

Scope extraction. The AI reads the plain-language description and pulls out the specific details that need to live in the job record: floor area, surface types, room breakdown, access constraints, product notes. It doesn't need the input to be formatted — it infers structure from natural language and maps it to the right fields in ServiceM8.

Pricing estimation. With the job type confirmed by the selector and the scope filled in by the AI, a preliminary price range is calculated against Pristine Properties' internal pricing rules. This isn't a binding quote — it's a structured estimate that gives the team a starting point before any site visit, and gives a client enough information to know whether to proceed.

Completeness checking. If the AI can't extract something it needs — the floor area is missing, the access situation is ambiguous — it flags the gap rather than guessing. The job record is created with a note for the team to confirm that specific detail on the follow-up call or site visit.

Escalation routing. Jobs that fall outside standard scope, involve compliance requirements, or contain details that warrant a senior review get flagged and routed to Basil directly. The AI doesn't try to handle what it shouldn't.

The whole pipeline takes under ten seconds from submission.

Step 3 — Everything Is Pre-Populated Before the First Call

Within seconds of submission, the structured data lands where it needs to be. The back office doesn't receive a raw email to interpret — it receives a job that's already been partially set up.

In ServiceM8, the job record exists with the correct template applied, the scope notes in the right fields, and the pricing estimate entered. If the AI had enough information to suggest an assignment, a technician is pre-allocated. If not, the record is ready for a coordinator to assign in seconds rather than minutes.

In Xero, a draft quote has been generated with the relevant line items. The team reviews and approves it — they don't build it from scratch.

If the enquiry came from a client rather than a team member, the client also receives a response — a personalised acknowledgement with the preliminary price range and a clear next step, sent automatically.

A site visit may still be needed before a final quote is confirmed. A back-office review still happens before anything is formally booked. What's changed is the quality of everything going into those steps. The team arrives at the site visit or the follow-up call with the job already scoped, the pricing already estimated, and the record already waiting in every system.

The Tool the Team Actually Uses

One of the clearest signals that this system works is how the Pristine Properties team uses it themselves. When a staff member takes a new enquiry over the phone, they open the quote tool and fill it in as they talk — selecting the options, typing the details into the free-text fields as the client describes them. By the end of the call, the quote capture is done. ServiceM8 and Xero are already updated.

This wasn't planned as a staff-facing tool. It was designed for inbound clients. But the interface turned out to be easier and faster than whatever the team was using before — and more consistent. No two staff members interpret a job differently when they're selecting from the same options and letting the AI structure the same free-text fields.

"The thing I didn't expect," Basil says, "was how much better our phone calls got. When we do need to call a client now, we already know the basics. The conversation is about confirming details, not gathering them from scratch."

The site visit still happens when it needs to. The back office still reviews and confirms before anything is finalised. What's changed is the quality of information flowing into those steps — and how much less time the team spends transferring that information between systems.

The Integration Layer

The reason this works without a custom-built backend is the integration infrastructure underneath it. The quote form is a lightweight web component embedded on the site. n8n handles the orchestration — receiving the form submission, passing it to the AI, parsing the response, and pushing the structured data to each connected system:

• ServiceM8 receives the job record with template, scope notes, and technician assignment
• Xero receives the draft quote with line items pre-populated
• Google Calendar receives a placeholder booking block if a time preference was specified
• Slack receives a formatted summary for the ops team
• The client's email receives the personalised response

No data is re-keyed. The same information the client typed into the form appears — correctly formatted, correctly categorised — in every system the business uses.

The Numbers

Since going live with the new system, Pristine Properties has seen:

• Average response time to new enquiries down from 6–8 hours to under 2 minutes
• Quote conversion rate increased, attributed to faster response and more professional quote presentation
• Admin time per new enquiry reduced by approximately 25 minutes — the time previously spent on scoping calls, manual quote creation, and CRM data entry
• Zero missed enquiries — every form submission is logged and tracked regardless of when it arrives

The system handles enquiries at 11pm on a Sunday the same way it handles them at 10am on a Tuesday.

Building This for Your Business

The Pristine Properties system isn't a product — it's a custom workflow built on tools that most Sydney service businesses already use or could easily adopt. The components (a web form, n8n, OpenAI, ServiceM8, Xero) are all accessible. What makes it work is the integration design: making sure the AI extracts the right information, mapping that information correctly to each downstream system, and building the right rules for when to automate and when to escalate.

At Proanalytica Technologies, we design and build these end-to-end systems for service businesses in Greater Sydney. If you're handling new client enquiries manually — phone calls, inbox triage, handwritten quotes — there's a version of this workflow that fits your business.

Get in touch to talk through what it would look like for you.