QUESTIONS
WE GET ASKED

It reads your documents and returns structured output. You feed it a blueprint, a freight invoice, a customs package, a set of daily site reports — and it extracts the relevant data, cross-references it against other documents or internal rules, and flags issues. The output is a formatted report your team can act on directly. It doesn't replace your project management system or TMS — it handles the document intelligence layer your team is currently doing manually.

Custom-built for your operation. Every system starts from a discovery session where we map the specific document types you work with — your blueprints, your freight invoices, your customs formats. The model is fine-tuned on your actual documents, not trained on generic data. A residential GC and an industrial contractor get different builds. A freight broker and a 3PL get different builds. There's no template being reskinned — it's built around what you actually do.

PDFs (including scanned), spreadsheets (Excel, CSV), Word documents, and plain text. Drawing sets, project manuals and specifications, RFI logs, scope of work documents, subcontractor quotes and exclusion letters, daily site reports, and subcontract agreements. If you have document types that aren't listed — put one through it during the free trial and you will see exactly what the system can extract from it.

It doesn't replace your ERP, TMS, or project management system. It doesn't connect to live data feeds or carrier APIs in real time. It doesn't answer open-ended questions or function as a general-purpose chatbot. It processes documents you feed it and returns structured findings. It's a document intelligence layer — not an all-in-one platform. If you're looking for a full operations suite, this isn't it.

A plan opens the five live programs in your browser for the number of seats it covers, on Marapone's own construction-trained model, with the reference data packs for your jurisdiction included. What you produce — takeoffs, estimates, levelled bids, reports — is yours, exports in formats that open in Excel, Bluebeam and a PDF reader, and stays yours after you cancel. What you do not get is the source code or the model weights: Marapone runs the engine, which is what keeps one flat price per firm possible instead of a five-figure implementation.

Because the programs are too heavy to run on your laptop and too expensive to hand over. Reading a 400-sheet drawing set takes real compute, and that compute costs money every month whether we charge for it or not. A subscription is the honest shape for that: you pay for what is being run for you, we keep improving it, and you stop paying the month you stop finding it useful. It also means everybody is on the current version — there is no customer stranded on a build from eighteen months ago because they bought once and never upgraded.

Four plans, all billed monthly or yearly. Solo — $49 CAD a month gives one seat on the AI Estimator, for the estimator who already has their quantities. Field — $99 CAD a month adds Blueprint Auditor, so the measuring happens here too and feeds the estimate with nothing retyped. Pro — $249 CAD a month opens all eight live programs for up to five people, wired together so one job runs from the drawing set through the awarded subcontract to close-out. Firm — $549 CAD a month covers up to twenty seats with a workspace per project, priority processing and a named contact — and on yearly billing it includes all five data packs as files, which stay yours even if you never resubscribe. The seat counts on Pro and Firm are ceilings, not requirements: one person can use either plan in full, at the same price. Pay yearly and two months are free. The same login also covers the mobile app when it lands in Q4 2026. Full details are on the Pricing page.

Alongside the plans sit two things you buy rather than rent: the data packs, from $59 — the reference libraries the programs read, sold as files with no software attached — and custom build work from $4,900, quoted once and then run for you under your plan. Pay for a pack and that money comes off your first plan invoice if you start one within 60 days.

A data pack is the reference library one of the apps reads, sold on its own as files — PDF, CSV and JSON. A code-check ruleset, an assembly schema, a jurisdiction rate table, a bid-levelling template or a master scope checklist. There is nothing to install and no licence key, because there is no software in the box.

The difference is who does the matching. With a pack you read your own drawings, specs and quotes against the library by hand. The app reads them for you and reports what it found, cited to the clause and the sheet. For a single job the file is often enough; the app earns its price when you are doing it every week.

Packs are $59–$99 each, or $349 for all five. A pack is files you download and keep, so it cannot be un-copied and is not refundable; the plans are not refunded either, you simply cancel any time from your own dashboard and keep access to the end of the period you have paid for.

The plan is the cost. Support, every update, unlimited use and every jurisdiction's data packs are included in it — there is no implementation fee, no per-seat AI add-on, no usage charge and no separate support retainer. Your bill is the same in a busy tender month as a quiet one.

Two things cost extra, and only if you want them: a custom build, if the eight programs genuinely do not cover your workflow, and the data packs if you want the reference libraries as files on their own.

To us. The programs run on Marapone-operated infrastructure, so your documents are uploaded to be read. There is no configuration that keeps processing on your own machines, and we would rather say that here than let you discover it later.

What we do promise: a workspace only your account reaches, no third-party AI provider anywhere in the path — the model is our own — nothing used to train anything, and export or deletion on demand. If an owner or architect has contractually barred putting tender drawings on a third-party service, a plan is the wrong purchase and we will say so rather than sell it to you. The one exception is deployment into your own cloud tenant, which is enterprise work priced as a project.

We have no access to your machine or your network — nothing is installed there, so there is nothing to reach. Your documents do sit in our infrastructure, because that is where they are read.

Access to them is restricted to the small number of people who operate the service, and used only for support you have asked for. They are never used to train a model, never shared with a third party, and deleted on written request.

They go into a workspace only your account reaches, and they are treated exactly the same as a paying subscriber's — a trial is a subscription that has not billed yet, not a different tier of privacy. We never use client documents for model training and never share them with third parties, and you can export or delete anything from your account whenever you like, including after you cancel. If you're concerned about sensitivity, the system still produces meaningful results from a partially redacted document in most cases.

Delete the project from your workspace and its documents, extracted data and generated reports go with it. Export anything you want to keep first — that is one click per report, and it is worth making a habit of at job close, because the export is the copy that outlives the subscription.

Because the programs run on our infrastructure, this is a real deletion rather than dragging a folder to the trash. If you want the whole account and everything in it removed, ask in writing and we will do it. If your document-retention policy requires an auditable purge procedure, we will write you one against your actual workspace.

ChatGPT is a general-purpose assistant: it does not know your drawing set, it will not cite the sheet a number came from, and it is happy to give you a confident answer it invented. Both are hosted services, so "it stays on your machine" is not the difference and we will not claim it is. The difference is what happens to the document: our programs are built for one job each, run on Marapone's own construction-trained model with no third-party AI vendor in the path, print the sheet and clause behind every finding, and return a structured file — a takeoff, a levelling matrix, a drafted RFI — rather than a paragraph. Nothing you upload trains any model.

No system catches everything — and we won't claim otherwise. What the system does is dramatically reduce what slips through compared to manual review, and do it consistently, without getting tired or distracted. The system is built as a first-pass intelligence layer, not a replacement for human judgment. Your team still reviews the output and makes the final call. The goal is that by the time something reaches your desk, the obvious noise is already gone and what's left is what actually needs a decision. We deliberately don't publish a headline accuracy percentage, because the honest number depends entirely on your document set — the free trial measures it on your own documents, before you commit to anything.

Support is in the plan, not sold beside it. Every plan includes email support — one business day on Pro, same day with a named contact on Firm — and every improvement lands the day it ships, with no version to buy and no upgrade to schedule. What is quoted separately is bespoke work: a new workflow built around your own documents, or an integration into Procore, Sage or your ERP. That is project work from $4,900, and it runs under your plan once it is built.

Those are project management platforms that added AI. Marapone is the AI. The differences that matter: (1) price is set by seats, not by your annual construction volume, and there is no implementation fee — Pro is $2,490 a year for up to five seats against $15,000–$40,000 for a comparable platform package; (2) your documents are read by our own model on our own hardware, never handed to a third-party AI vendor and never used for training; (3) the model is built for construction documents specifically — drawing sets, project manuals, sub quotes — rather than being a general assistant bolted onto a file store. If you already have Procore and it's working for project management, Marapone slots in alongside it as the document-intelligence layer (we integrate with Procore via API or file export). See the detailed side-by-side comparisons: vs Autodesk Construction Cloud, vs Document Crunch.

No. The five live programs run on Marapone's hardware and arrive in a browser tab, so whatever your estimating team already opens email on is enough — Windows, macOS, Linux, a site laptop, a borrowed one. No GPU, no server, no minimum spec, nothing for IT to approve. The only requirement is a connection. Custom builds that integrate with your own systems are the exception, and any infrastructure they need is scoped and quoted before you commit to anything.

The five live programs are finished and running — you sign in and use them the same day, with nothing to install. Custom build work is the part with a schedule: one to three weeks from signed scope to the program being live in your account, depending on complexity. A customised workflow typically runs one to two weeks; an enterprise integration runs two to three, longer where a security review or an undocumented API is in the way. The timeline starts after the scoping call and written specification are agreed on — the scoping phase before that runs separately and typically takes three to five business days total.

Yes — that's the point. There is nothing for IT to install, provision or approve beyond the login. You sign in, upload a drawing set and the programs are open; most people are running their first real set the same afternoon.

If it helps, we will walk your team through it on a call: where documents go, how to read what comes back, which export lands in your system, and what to do when something looks off. The programs are built to be used by whoever already handles your document workflow, not by a developer.

There is no update path to manage, because there is no version on your side. Improvements reach you the day they ship, at no extra cost and with nothing to download — you cannot fall behind, and there is never a new edition to buy.

What ages is content rather than code: fee schedules change, rate tables drift, tax and statutory numbers get amended. Those live in the JSON data packs and get republished as the jurisdictions move, on every plan including Field.

The honest half of the answer is the other direction. Because we run the engine, "will this still run in three years" depends on us still being here and you still being subscribed. That is the real trade, and it is why everything you produce exports on demand in Excel, CSV and PDF, why the data packs are sold separately as files, and why there is no annual commitment required to get the good price. If we disappear, your work does not.

Fixing anything that does not work as described is not a support tier, it is the plan. There is no warranty window to run out and no defect you have to pay to have corrected.

The fair question underneath this one is what happens to you if Marapone goes away, and we would rather answer it plainly than pretend it cannot happen. Everything you produce exports on demand — takeoffs, estimates, levelled bids, reports — in formats that open in Excel, Bluebeam and a PDF reader, and you can export the lot at any time without asking us. The data packs your programs read are sold separately as files precisely so the reference layer survives us. What you would lose is the programs themselves, and we are not going to dress that up: that is the trade you make when the engine runs on somebody else's hardware, and it is the reason we do not ask for an annual commitment to get the good price.

Bespoke work is quoted per engagement rather than as a published tier, because the useful version depends entirely on what you want changed.

That's what the 7-day free trial is for. Make an account, add a card at Stripe, and the Field plan opens on your own drawings for a week — unlimited runs, no document cap, no AI credits, no demo data. You see exactly what the programs extract and flag from your own documents before a dollar changes hands. If it doesn't produce something useful, cancel before day 8 and you've spent nothing finding that out.

We're based in Canada and Italy, and most of our clients are in those markets — but the work is not geographically restricted. We've scoped builds remotely and can ship hardware internationally when needed.

The more useful answer is about the software rather than us. Rates, tax, statutory holdback, permit fees and code references aren't compiled into the products — they load from JSON data packs that ship in your source repo, and the scope knowledge base is indexed by CSI MasterFormat rather than anything Ontario-specific. Your market's pack is built for you and you can edit it yourself afterwards; Toronto is not a default you have to opt out of, it is simply the pack that is already finished. What we don't have is a shelf of pre-built international packs ready to license — the regions page is explicit about where each market stands.

We intentionally keep our active build count small — typically two to three projects at a time. We're a boutique operation, not a software factory. This isn't a limitation we're trying to scale past — it's how we ensure every build gets the attention it requires. If our pipeline is full when you reach out, we'll tell you directly and give you an honest timeline.

The 7-day free trial. Create an account, add a card at Stripe, and the Field plan opens immediately — Blueprint Auditor and the AI Estimator, on your own jobs, unlimited. No sales call, no pitch deck, and nothing charged until day 8. If it earns its place, do nothing and it carries on. If it doesn't, cancel in one click and you've spent nothing.