Modern GBQ Platform Overview

Modern GBQ is developed by GBQ Australia as a governed, multi-sided platform for residential construction. It connects homeowners and developers who need a trustworthy estimate, builders who want work that arrives already scoped, and suppliers whose published prices become the figures inside real estimates — with consultants and the platform’s own operations team working alongside them. At its core is the Grace Estimation Method, a proprietary, governed set of estimating formulas that turns a project description into engineered quantities, line by line, with the working visible.

Built around

the Grace Estimation Method (GEM) — governed formulas with visible working

Who it serves

homeowners & developers · builders · suppliers · consultants

Markets from day one

Australia (English) and Egypt (Arabic), with a configurable multi-country core

Stage

V1 preview — the full stack is live for review ahead of launch

Residential construction begins with a number nobody can check. A homeowner collects quotes that price different scopes, built on different assumptions, none of them showing their working; a builder prices a sketch and a hope, and pads for the uncertainty; a supplier’s price only enters the picture after the builder has chosen it, by which time the estimate is fixed. Every party is negotiating around a figure none of them can trace.

Modern GBQ inverts that order. The quantities come first, computed by governed formulas from the project’s geometry and choices; the prices attach to the quantities, each line recording which offer or rate produced it and why; and the builders respond to that scope — the same scope, for everyone invited — so the comparison is like-for-like and the homeowner reads it blind. The estimate becomes a shared, auditable object rather than a guess.

1. The project is described — the site (looked up by address and pinned on a map), buildings, storeys and rooms, the construction system and finishing quality — in a workspace that also keeps the people involved (owner, architect, estimator, project manager and others, each with a defined role).
2. GEM computes the quantities. The Foundation Pack’s 64 formulas run over the project in dependency order and produce every quantity the estimate needs, each with its inputs, its formula and its confidence level shown.
3. The quote is priced, line by line, from published supplier offers (or recent market prices where none exists), trade labour rates for the site’s locality and the market’s tax — with the basis of every figure recorded on the line.
4. The homeowner reviews and approves a version. Versions are never edited in place; a re-price creates the next one and the history stays intact.
5. Builders are invited to respond to the approved scope, within a five-day window. Responses are compared blind — figures, programme and qualifications before any name — until the homeowner records their feedback.
6. A preferred builder is selected and notified, acknowledges, and the two parties conclude their contract directly. The platform’s part ends there, by design.

GEM is a catalogue of governed formulas, not a black box. The Foundation Pack ships 64 formulas across five families — Geometry, Structure, Finishes, Openings, and Mechanical, electrical & plumbing — each declaring the rules it depends on, the master item it measures and a confidence level (Derived, Preliminary, Benchmark or User input). A calculation runs the graph in dependency order and reports, for every line, the quantity, the inputs it used and the expression it evaluated.

The pack is calibrated against a documented sample project — a 240 m², two-level, four-bedroom, three-bathroom house — for which it reproduces the reference figures of 24.16 m³ of concrete and 2,658 kg of reinforcement. That sample is built into the GEM workbench, so any change to a formula can be checked against the same reference before it is published.

Every formula and coefficient is a governed record: it moves Draft → In review → Approved → Published, can be superseded by a new version or retired, and the audit trail records who did what and why. A quote is priced against the versions published at the time, and that snapshot is kept with the quote — so an estimate produced today can be explained, unchanged, years later.

Pricing is deterministic, and every line says what priced it. For each material line the platform resolves a product variant, then looks for a published supplier offer for it in the quote’s market and currency, valid today — preferring a supplier’s Preferred offer, then the lowest unit price, then the most recently updated. Where no offer exists it falls back, on every pathway, to a recent market price — a supplier’s confirmed price ahead of one the sourcing agent found, newest first, each dated and linked to its source — then to an item-scoped indicative rate, then a category average, and finally carries the line at nil and lists it. Pricing never waits on a supplier — a quote is always priceable, and the supplier whose product was chosen is simply notified. What differs between pathways is the input:

Labour is priced separately for each line — from the master item’s installation activity to a trade rate for the site’s locality, with a project-condition factor (Easy, Standard or Difficult) — and tax applies once, at the market’s published rate. Material, labour and tax are shown as separate figures, and each line carries its price source: Supplier offer (naming the offer and the supplier), Market price (dated, with its source), Indicative item rate, Category average or Not priced. Nothing is silently assumed.

An engagement starts from a builder-ready quote — an approved Detailed quote, or an approved Design quote for a project with architectural drawings on file; a Ballpark figure can never go to tender. Invited builders receive the exact approved scope, including any lines the homeowner excluded, together with the drawings, site plan and specifications the project team has chosen to share (each project document is marked shared or private; drawings, site plans and specifications are shared by default, reports and photos are private by default), and respond with one of four answers (Accept, Revised proposal, Request discussion or Decline), a price revision and six written fields covering programme, qualifications, exclusions and proposed alternatives. Responses can be revised while the round is open; every version is kept.

The homeowner’s comparison report is anonymous first: responses appear as Builder A, B, C… with every figure visible and no names. Recording feedback closes the round and reveals identities; only then can a preferred builder be selected and notified. The rules that keep the round fair are enforced by the platform, not left to discretion — the response window, the rounds allowed per quote and the cooldown between rounds are governed Marketplace rules in the Configuration Centre, versioned and approved like every other setting:

• A limited number of rounds per quote (two by default), and a further round only after a cooldown following the previous comparison report (five days by default) — a quote cannot be shopped around indefinitely.
• Written reasons are required to decline a selection or withdraw one, and are kept on the record.
• A builder rating starts at 5.00 stars; the one thing that lowers it is declining after being selected, by half a star each time, with the reason shown on the builder’s history. Declining an invitation costs nothing.
• Verification before the marketplace. A builder or supplier organisation is created Pending and is verified — or rejected, with a reason its members can read — by the platform’s team, never self-declared. Only verified builders are listed and invitable; only verified suppliers can record offers and prices. Homeowners run their own sourcing without any staff step, and each person has one homeowner organisation.
• Symmetry of anonymity: during the blind phase the homeowner sees no builder names, and builders see the site’s locality but not the homeowner’s identity or street address.

The catalogue is a governed hierarchy: master items (what an estimate can contain, in a fixed unit of measure) → products → product variants (what a supplier actually sells) → supplier offers (a price for one variant in one market and currency, with validity dates and terms). Suppliers manage their offers in a workspace that surfaces lapsed and expiring prices before they cost visibility, extend validity in one click, and maintain whole catalogues through a CSV export–edit–import round trip with a dry run that reports what every row would do before anything is written.

Every product can carry photos — a primary image and a gallery, for the whole product or for one variant. The product’s photo appears beside its name on the quote wizard’s finishing selections and beside every alternative a homeowner can swap to, so a tile, a cladding board or a tapware range is chosen by what it looks like rather than by a code. Photos are uploaded by the catalogue team on the product’s master-item screen and can be hidden again without being deleted.

To let the platform price standard Australian projects from the first day, an Australian starter catalogue ships with the platform: 274 master items covering the register the formulas measure against, a generic product and variant for each of its 190 material items, 188 item-scoped indicative rates in AUD (approximate market figures, meant to be replaced as real prices arrive), and 760 finishing defaults across four quality levels. Indicative rates are the fallback behind real published offers on every pathway — a line priced from one says so, and a published supplier offer takes precedence the moment it exists.

The platform’s material sourcing agent finds current public listings for building materials and records each one as a dated market price against the catalogue — matched to an existing product, or to a new Draft product carrying the listing’s specifications when only a generic material type matches. Every record keeps the source listing’s link, the capture date and a match-confidence score. A market price is a fallback, not a commitment: it prices a quote line only when no published supplier offer exists, a supplier’s own confirmed price always outranks one the agent found, and the line says which it used. Where a reviewer names a supplier, the agent also proposes matching listings as Draft offers that a person reviews — least-certain matches first — before they can be published. Listings the agent cannot place in the catalogue are reported, not invented; duplicates are skipped; and total model spend is held under a configurable monthly cap that the agent checks before every run.

Market prices are kept current three ways, so the figures behind an estimate never quietly age: the operations team saves searches as schedules the agent re-runs on a cadence (weekly by default, within the same spend cap); suppliers confirm, correct or record the prices for their own products from their workspace, and a supplier’s figure outranks the agent’s; and a whole price list can be uploaded as a CSV — by a supplier for their range, or by the operations team as a market list — with a dry run first. A price older than the configured window (60 days by default) drops out of pricing altogether.

Construction knowledge is configuration, not code. The Configuration Centre governs fourteen domains — Countries, Currencies, Regions, Price regions, Taxes, Languages, Construction systems, Reference types, Quality packages, Quote Wizard steps, Notification templates, Marketplace rules, GEM formulas and GEM coefficients — through one lifecycle: a record is drafted, submitted with a change reason, approved, and published with an effective date; a new version supersedes the old one, and every step is on the audit trail. Adding a country, a tax or a formula family is a governed change made by the operations team, not a software release.

The GEM workbench sits on the formula domain: a family filter and dependency chains for reading the graph, a searchable picker for targeting formulas at master items, a test calculation pre-filled with the sample project — able to preview unpublished versions — and a bulk submit-approve-publish loop for releasing a pack.

• Tenant isolation. Every record belongs to a tenant, and every query is scoped to it at the data layer, not left to each screen.
• Role-based access. Fifty-nine distinct permissions, grouped into roles that the platform’s administrators assign; the API refuses what a role does not grant and records each refusal as a security event.
• Two-factor sign-in with an authenticator app, recovery codes, and a login flow that asks for the code rather than failing.
• An audit trail on governed records, security events and marketplace actions — who, what, when and why.
• Legal documents in the product. Ten documents — Terms of Service, Marketplace Agreement, Privacy Policy, Billing & Refunds and Dispute Resolution, for Australia in English and for Egypt in Arabic — are readable without an account, accepted at registration and organisation creation, and recorded with the version accepted; a re-versioned document automatically falls due again. They are prepared for solicitor review ahead of launch.
• The neutral boundary is enforced in software. The platform holds no payment rails, no escrow and no contract signing; the screens that could be mistaken for a transaction carry the same notice as the Terms.

The commercial model combines recurring software revenue in V1 with marketplace transaction revenue planned for V2 — kept separate by design so each can be switched on safely, transparently and market by market.

Three commitments govern the model. Project payments stay off the platform in V1 — the architecture and feature flags for V2 settlement exist, but processing remains disabled until security, legal, tax, regulatory and operational approval is in place for a given market. Transaction volume is never reported as revenue. And sponsored placement is labelled and stored separately from objective technical or suitability scoring, so a promoted listing can never masquerade as a recommendation.

The V1 preview is a complete, running stack — API, database and web application — deployed for review, with the full homeowner journey working end to end. Available today:

• Accounts, organisations of every type and their members, with two-factor sign-in and in-product legal consent.
• The project workspace with participants, roles and the project’s drawings and documents (shared with invited builders or kept private); the Quote Wizard with its three pathways, GEM calculation, finishing selections with product photos, per-line material and labour pricing, tax, versioning and approval.
• The builder marketplace: invitations carrying the priced scope and the shared drawings, responses and revisions, the blind comparison, feedback and reveal, selection, notification, acknowledgement, decline and withdrawal with reasons, re-selection, round limits and ratings.
• Builder and supplier workspaces; the supplier catalogue with product photos, offers, validity management and the CSV round trip; the sourcing agent with its market-price records, saved schedules and the AI-sourced proposal review queue; supplier confirmation of market prices and price-list upload.
• The Configuration Centre across all fourteen domains, the GEM workbench, the catalogue screens and the Australian starter pack.
• Notifications (in-app), support tickets with priorities, an administration console for roles, permissions, users and organisation verification, and these public help pages.

Engineering discipline behind it: a modular monolith with a single authoritative database, more than five hundred automated unit tests plus integration tests against a real database, and a continuous-integration pipeline that validates every change and can deploy to cloud infrastructure defined as code. Planned and not yet available: email and SMS delivery of notifications, reading uploaded drawings into the estimate automatically, project geometry entered directly in the estimate wizard, an in-app two-factor enrolment screen, and a native mobile experience.

Before public launch, the remaining steps are external rather than engineering: standing up the production cloud environment, connecting the email provider, solicitor review of the ten legal documents, and a pilot with real projects to calibrate the GEM formulas.

Candidates already identified for the first releases after launch, from the product’s design work: a line-level quote builder, subcontractor management, invoicing and progress claims, contract and variation tracking, budget-versus-actual reporting, and client messaging.

What can be examined today is a governed, working platform rather than a deck: the accepted architecture decisions that fix its commercial boundaries, a stable business-rule catalogue, feature flags separating released, entitled and country-approved capabilities, immutable quote and GEM snapshots, separate subscription and marketplace ledgers, and the code, tests and deployment pipeline behind all of it. A walkthrough of the live preview and the governance pack are available on request.

Equally, what is not yet prepared is stated plainly. The following diligence items are being developed separately and are not represented on this page: verified market sizing and competitor analysis; a financial model and five-year forecasts; customer and supplier traction evidence; pricing validation and unit economics; regulatory advice for each V2 payment market; IP legal review and ownership confirmation; the data-protection and terms-of-service review; and payment-provider commercial proposals.

Investor enquiries are welcome through GBQ Australia; registered users can also reach the team through the platform’s Support page.

• Every figure the platform produces is an estimate prepared with the Grace Estimation Method — not a quotation, and not an offer from any builder or supplier.
• Modern GBQ does not process payments, hold funds, or create contracts. Selecting a builder is an expression of intent; the contract is concluded directly between the parties.
• Builder comparisons are blind first; identities are revealed only when the homeowner records feedback.
• Nothing prices a quote from a draft, and nothing an AI proposes goes live without human review.
• Every legal document is readable at the Legal page without an account, and acceptance is recorded with the version accepted.