Marapone AI Estimator New

Takeoff measures. This prices. AI Estimator — construction cost estimating software for general contractors in North America, Europe and Australia

The Blueprint Auditor tells you there are 24,000 square feet of partition. The AI Estimator turns that into studs, board, tape, fasteners, waste and crew-hours, then adds the risk, the contingency, the municipal soft costs and the margin you can actually win at. Opens in a browser on any machine — Windows, Mac or Linux. From $69 a month.

Open it on Solo — $69/mo →

See every screen

Or book a walkthrough on one of your own jobs first → Or download the one-page spec & price sheet (PDF) ↓

✓ Opens in a browser ✓ No third-party AI vendor ✓ Stands alone or pairs with takeoff ✓ Any modern browser · mobile Q4 2026

The real app, recorded · a 38-unit Toronto infill priced end to end

See every screen

Or book a walkthrough on one of your own jobs first → Or download the one-page spec & price sheet (PDF) ↓

✓ Opens in a browser ✓ No third-party AI vendor ✓ Stands alone or pairs with takeoff ✓ Any modern browser · mobile Q4 2026

57

Parametric assemblies

Across 20 CSI divisions

27

Risk rules

Each with a cited reason

25

Requirement types read

From specs, ITBs and addenda

4,000

Monte Carlo iterations

Per cost distribution

1.0 Two tools, one pipeline

This is not a takeoff tool.

Takeoff and estimating get sold as one thing and they are not. Measuring is a geometry problem. Pricing is a judgement problem. We built them as two products so you can buy the one you are missing, or wire both together and never retype a quantity again.

Upstream

Blueprint Auditor

Reads the drawing set and answers how much is there.

→ Quantities measured off the sheets
→ Scope gaps and drawing conflicts
→ Code and tender risk flags

See the auditor →

Downstream

AI Estimator

Takes those quantities and answers what should we bid.

→ Assemblies, waste factors and crew-hours
→ Contingency sized to the actual risk
→ Permits, DCs, HST and the margin call

You are here

Either one works on its own. Already have takeoff you trust? Import a CSV or Excel sheet of quantities and the estimator never needs the auditor. Already price by hand and only want the measuring done? Buy the auditor alone. Most teams end up running both, because the seam between them is where quantities get retyped and errors get born.

2.0 Inside the app

Every screen, no mock-ups.

Everything below is the shipping application priced against the demo project that ships with it: Lakeshore Yards, the four-storey, 38-unit residential infill at 118 Jefferson Avenue, Toronto that this site uses throughout — 42,000 SF, 15 months, constrained site, winter start. The figures come from that fixture so you can check them against what you see here; the assemblies, the rate tables and the risk rules are the ones your own job goes through.

01 — Risk & contingency

The number under the number.

27 rules across scope, schedule, market, site, regulatory, design and financial exposure. Each one that fires gets a written reason, a dollar exposure and a mitigation, not a colour-coded dot.

On this job the model wants 9.0% contingency. The bid carries 5.0%. That gap is $362,535 of unpriced exposure, and it is on screen before the bid goes out, while it is still worth knowing.

Contingency clamped into a band by design stage, so a pile of small flags cannot produce an absurd number
Confidence scored by division — you can see which trades are assumption and which are quote
Every rule cites the code, by-law or statute it comes from

02 — Bid strategy

Where to put the margin, and why.

Four numbers every estimator has in their head and almost never writes down: the walk-away price, the competitive price, the recommended price, and the stretch. Here they are calculated, and the win probability behind them comes from your own recorded outcomes.

Margin ladder with expected value at every rung — the pure EV optimum is often below where you want to be
A risk floor that refuses to let margin cover a risk that was never priced
Priced value-engineering options for when the number comes back too high
Record the outcome afterwards and the curve sharpens on the next bid

03 — Municipal soft costs

The line item that dwarfs the finishes.

On this project the development charges alone are $3,153,979 gross — reduced to $1,576,989 under the Canada–Ontario Development Charge Reduction Program. That is more than the entire interiors package, and generic estimating software does not know it exists.

City of Toronto 2026 fee schedule — area rates by occupancy, 4.5% Construction Surcharge, Building Code Act administration fee
Development charges gross and reduced side by side, with demolition credits
Parkland, Community Benefits Charge, HST at 13% and the new housing rebate
Every levy tagged CONTRACTOR or OWNER, so nothing lands in the bid that should not be there

Municipal schedules change annually. Every figure is editable in-app and stored in your own folder, no dependency on us to stay current.

04 — Document intelligence

The clauses that cost money.

The takeoff reads the drawings. This reads everything else that binds the price — specifications, instructions to bidders, addenda, geotechnical reports, emailed clarifications, and pulls out the 25 requirement types that actually carry cost or risk. Three documents on this job produced 26 requirements.

Bonding, insurance, liquidated damages, holdback and prompt payment
Fair wage, phasing, restricted hours, road occupancy, sole-source products
Toronto Green Standard, green roof, heritage, tree protection, excess soil
Each finding shows the excerpt it came from, and converts to a qualification, a risk flag or a markup change with one click

05 — Learning from your jobs

It stops using averages and starts using your crews.

Feed in completed jobs with actual costs and hours. The catalog defaults get replaced by what your people actually do: drywall running 14% over, carpentry 10% under, finishes consistently 7% hot. Those factors then apply to every future estimate automatically.

Factors are shrunk toward 1.00 by sample size — two jobs get a quarter of the signal, eight or more get all of it
One bad job never permanently distorts your pricing
Learned at assembly, trade and division level, whichever has the evidence
Import from CSV, or add a job by hand in a minute

06 — What-if & Monte Carlo

Eight preset stress tests, custom levers, tornado sensitivity, and a probabilistic cost distribution that answers the only question that matters at 2 a.m.: what is the chance this job loses money. Scenarios run against a copy — nothing moves until you apply one.

07 — The estimate itself

Every assembly expanded to its components, grouped by CSI division, each line traceable back to the catalog entry, your rate sheet or the drawing it was measured from. Override any price and it locks — repricing will not quietly overwrite a number you know to be true.

3.0 Nothing redrawn for the website

Every screen, at full resolution.

Screenshots straight out of AI Estimator. Click any one to open it at capture size and read the actual figures in the actual cells.

The estimate at a glance — direct cost, bid total and where the margin sits

The priced estimate, assembly by assembly across 20 CSI divisions

The risk register — each risk costed, with the rule that fired it

The margin ladder and win probability, from your own past bids

Permits, development charges and taxes for the jurisdiction

Scenarios side by side, so the walk-away number is explicit

The rate tables — editable JSON, not a locked database

Click to enlarge · press Esc to close

4.0 Your jurisdiction, properly

Your city, not "North America".

Generic estimating software is accurate everywhere and therefore precise nowhere. This one runs tuned to a single named jurisdiction — any Canadian province, US state or European country, because the city is where the money actually leaks, and a province-wide average hides exactly the charges that decide a bid.

Every number below lives in an editable JSON file, with its citation attached. Toronto and Ontario are shown here because a finished pack is the only honest way to demonstrate the depth is real — that is the level of detail your jurisdiction's pack is built to, and it is built as part of your engagement.

Worked example — the Toronto & Ontario pack, shipped and in production

Building code

2024 Ontario Building Code (O. Reg. 163/24, in force 1 Jan 2025), SB-12 Zone 6 energy packages, 1.2 m frost depth, AODA barrier-free requirements.

Payment & statute

Construction Act (Ontario) — 10% statutory holdback, 28-day prompt payment, s. 85.1 bonding thresholds on public work.

Permits

City of Toronto 2026 fee schedule — $214.79 minimum permit, $92.79 hourly examination, area rates by occupancy, 4.5% Construction Surcharge, $3.00/m² Building Code Act administration fee.

Development charges

Canada–Ontario Development Charge Reduction Program (40–60% off, from March 2026), demolition credits, gross and reduced shown side by side so the exposure if eligibility fails is never hidden.

Tax

HST at 13%, shown separately and never buried in a unit rate, with input tax credit treatment and the federal plus Ontario new housing rebate.

Municipal

Toronto Green Standard v4, green roof by-law Ch. 492, ravine and tree protection Ch. 658/813, Ontario Heritage Act, and excess soil under O. Reg. 406/19.

Schedule

Noise by-law Ch. 591 — durations built on the roughly 65 productive hours a week Toronto actually permits, against a seven-day assumption.

Labour

GTA labour benchmarks and market conditions as at mid-2026, used to validate imported rates and drive escalation — flagged for a second look and left for you to decide.

Building somewhere else?

None of the above is compiled into the engine. Tax, holdback, permit fees, code references and winter productivity are sections of a jurisdiction file; the 95-item material catalogue, 24 crew compositions and the city cost index are separate files again. Swapping market is a data edit, not a rebuild, and it is an edit you can make yourself.

How regions & rate packs work →

5.0 MaraponeAI

The model never decides a number.

This is the part most AI estimating tools get backwards. Prices come from the catalog and your rate sheet. Factors come from your job history. Contingency comes from the risk rules. The bid range comes from the strategy engine. All of it is deterministic and all of it is traceable.

MaraponeAI explains, summarises and drafts. It does not price. Every figure it quotes traces back to the audit trail, which is what makes the estimate defensible to an owner, and what stops a 30-year estimator from writing the whole thing off as a black box.

Even the conversational layer is deterministic. Type tighten drywall productivity 8% and show me the impact and it is parsed by rule, previewed before it moves a cent, and logged in the change history. An estimator typing an instruction about money needs it executed exactly — a misread intent that silently shifts a bid is far worse than an honest "I did not understand that".

Runs on any machine you own

A program that opens in any browser on Windows, macOS or Linux — five years old, borrowed, or the site laptop, with the mobile app arriving Q4 2026. Nothing to install, no admin rights, no hardware to provision. The reasoning runs on our GPUs, so yours is irrelevant.

Fully functional without a model

Every analysis, recommendation and explanation is available from the deterministic reasoning engine alone. Without a model you lose free-form narrative — you do not lose a single number.

Your rate work is yours to take

Rate overrides, fee-schedule edits and the job history the margin ladder learns from sit in your workspace, and every one of them exports to Excel and JSON on demand. We ship changes continuously without touching them, and if you cancel, the export is still the export.

Built by the people who made this page

The app and this website are the same design language on purpose — the same orange, the same mono type, the same dark plate. Estimating software usually looks like it was assembled by a different company than the one selling it. Ours doesn't, because it wasn't.

Exports into what you already run

Client proposal PDF, internal bid review PDF, Excel workbook, and CSVs for Procore, Sage, QuickBooks and Bluebeam, plus a buyout list. Client-facing and internal documents are deliberately separate — handing a client your margin ladder is a bad afternoon.

6.0 From drawings to submitted bid

Eight steps, and the last one matters most.

01

New estimate

Project, postal code, type, area, duration, site conditions.

02

Import quantities

Blueprint Auditor JSON, or any CSV/XLSX. Metric converts automatically. Anything it cannot confidently match is surfaced for you to resolve.

03

Import your rates

Your rate sheet becomes the primary source. Rates well outside your market's benchmark get flagged for review.

04

Analyse documents

Findings become qualifications, risk flags and markup changes you apply with one click.

05

Review the risk

Risk register, contingency gap, Monte Carlo, sensitivity.

06

Position the bid

Margin ladder, win probability, value engineering, walk-away number.

07

Export

Proposal PDF, internal review PDF, Excel workbook, integration CSVs, buyout list.

08

Close the loop

Record the outcome and the actual costs. The next estimate is measurably better because of it — this is the step everyone skips and the only one that compounds.

7.0 The output, not a screenshot of it

Read the bid it wrote.

Both documents below came out of the app on the Lakeshore Yards job you have been looking at, no editing, no cleanup. They are the fastest way to judge whether this thing actually understands construction. Open them before you spend anything.

Client-facing

The proposal

What the owner sees: the stipulated price, the cost build-up by division, inclusions, qualifications and exclusions, and a cover letter. No margin, no risk register — handing a client your margin ladder is a bad afternoon.

Open the PDF →

Internal only

The bid review

What you see before you commit: where this bid is weak, the $362,535 contingency gap, the recommended margin and win probability, the walk-away number, and the local specifics that move the schedule. This is the document that stops a bad bid going out.

Open the PDF →

Also exports an Excel workbook, a buyout list, and CSVs for Procore, Sage, QuickBooks and Bluebeam.

8.0 What a plan actually opens

A finished program, open in a browser tab.

One flat price per firm. No per-file charges, no metering, no implementation fee. The working program the minute you sign in, and you can cancel from your dashboard the month it stops earning its keep.

The arithmetic

On the demo job the risk engine wanted 9.0% contingency against the 5.0% the bid carried — $362,535 of unpriced exposure, on screen before the bid went out instead of in the post-mortem.

The Solo plan is $69 a month and prices every bid after that one. One tender where it moves the margin call, or catches a division priced on an assumption, has already paid for it several hundred times over.

Included in every plan

The complete program the moment you sign in — nothing to install, no licence key, no machine to provision
Unlimited bids, projects and documents on your plan's seats. No per-project or per-file fee, ever
All 57 assemblies across 20 CSI divisions, 27 risk rules, the Monte Carlo and sensitivity engines and the margin ladder
Every rate and fee schedule editable in-app — your labour rates, your productivity factors, your municipal charges
Proposal PDF, internal review PDF, Excel workbook and Procore / Sage / QuickBooks / Bluebeam exports
Every improvement the day it ships, no version to buy, no upgrade to schedule
Cancel any month from your dashboard, no long-term contract — see the plans

Not included — that’s a customised setup

From $1,500

An assembly catalogue rebuilt around how your company builds, or divisions you self-perform and price your own way
A soft-cost and by-law layer rebuilt to one municipality’s own schedule, beyond the jurisdiction pack
A live link into Procore, Sage, your ERP or your own cost database (the exports are files)
Bespoke development beyond the eight programs — that is quoted as project work

Buying the product first doesn’t lock you out of a build later — most teams start here, then commission the parts that are specific to them. See custom builds →

9.0 Separately, or together

Buy the piece you're missing.

Both open on a plan — the Estimator on its own from $69 on Solo, the pair on Field, all eight on Pro. No per-file charges and no metering. Full pricing lives on the pricing page; this is how the two programs fit together.

Option A

Estimator alone

You already have takeoff you trust — a measuring service, Bluebeam, or an estimator with a scale rule. Import a CSV of quantities and get the pricing, risk, soft costs and strategy layer on top.

Best for: estimators who measure fine and price by gut

Option B

Auditor alone

Your pricing is dialled in and the bottleneck is measuring and reviewing drawings. Run the Blueprint Auditor for takeoff, scope gaps and tender risk, and keep pricing the way you always have.

Best for: teams drowning in drawing review

Option C — most teams

The whole suite, wired together

Quantities flow from the drawings into the priced bid, SpecChecker catches the contradictions between the manual and the drawings before anything is ordered, the Bid Leveler normalises what the subs quote back, and its levelled position becomes the budget ScopeGuard measures every letter against. The seams between measuring, pricing and buying out are where errors get born — this removes them.

All eight on Pro · $499/mo, up to five seats

See pricing → Compare with the Blueprint Auditor

Which plan opens the AI Estimator?

Every plan. Solo opens it on its own at $69 a month; Field ($149) adds Blueprint Auditor for the takeoff; Crew adds SpecChecker and ScopeGuard; Pro and Firm open all eight programs. The five plans.

Is there a free trial?

Yes — seven days on Solo, or fourteen on Field, which includes it too. Either takes a card at Stripe and charges nothing until the trial ends; cancel before then and you pay nothing. One trial per account.

Does the AI decide the price?

No. Prices come from the catalogue and your rate sheet, factors from your job history, contingency from the risk rules and the bid range from the strategy engine — all deterministic and traceable. MaraponeAI explains, summarises and drafts; it does not price.

Can I use my own quantities?

Yes. Import a CSV of quantities from any takeoff you trust — a measuring service, Bluebeam or a scale rule — and get the pricing, risk, soft costs and strategy layer on top.

Where do my documents go?

To a workspace scoped to your account, on Marapone’s own infrastructure, read by our own model. No third-party AI vendor is in the path and nothing you upload trains any model. Cancel from your dashboard with no notice period, and your work exports to PDF, Excel and CSV.

Price your next bid with it.

Send us a real job — quantities, drawings, or just the tender documents. We will run it through the estimator and walk you through what it found, line by line, before you spend anything.

Open it on Solo → Book a walkthrough first

7 days free on SoloNothing charged until the trial ends.
Cancel from your dashboardNo contract, notice period or fee.
Your work exportsPDF, Excel and CSV — kept 90 days after you leave.
Nothing to installIt opens in the browser you already have.

11.0 For the person who has to sign off on it

The technical detail, before you ask for it.

Every claim on this page has a page behind it. No form, no call — read it now.