Skip to content
Documentation

Bid Leveler, in order.

Level a dozen subcontractor quotes to one scope, and see the whole shelf at once.

Bid Leveler

Level a dozen subcontractor quotes to one scope, and see the whole shelf at once. Included from Pro upwards.

1. Start with the board

Every package you are chasing, grouped by when it closes, with what each one is waiting for written out. It is the one screen that works with nothing open.

2. Import the quotes

PDFs as they arrived. What each one covers is read against the published scope checklist for that trade.

3. Read the matrix

Every scope element against every bidder: covered, excluded, or not mentioned — which is the one that costs money later.

4. Level and award

Adjustments for what a quote leaves out, the levelled cost beside the bid, and a clarification letter per bidder.

What it will not do

  • The scope checklists are sourced for Toronto and the GTA. Elsewhere they are a starting list.
  • A quote is read as written. A number nobody wrote down cannot be levelled.
  • The recommendation is a recommendation. Awarding is yours.

On a paid plan every program opens a sample set, so none of this has to be read before trying it. A free trial starts blank, on your own drawings. See the sample project.

In detail

In short

Tender a trade and a dozen quotes come back in a dozen formats — a formal letter with an exclusions schedule, a wall of prose with the carve-outs buried mid-sentence, a spreadsheet, a four-line email, a photograph of a fax. Levelling them by hand is a day in Excel, and the day usually ends with an estimator hoping they didn't miss anything. Bid Leveler reads all of them, normalises them to the same scope, and tells you what each bid would actually cost. Opens in a browser on any machine. From $499 a month.

A day in Excel, and it ends in a hope.

Five bidders quote the same package and none of them quote the same scope. One carries crane hoisting; one excludes it; two never mention it; the fifth prices it as an alternate. Put those five numbers in a column and the column is meaningless.

On the demo package the low bid was $1,142,000. Levelled, that bidder is not the cheapest — awarding on the bid form alone would have cost $84,986, and $226,160 of scope sits outside the bid numbers entirely whoever you award to.

Silence is the expensive part. An exclusion is on the record and can be priced. Silence is what reaches site unresolved, so the two are levelled the same way and flagged differently.

The grey bar is what they bid. The orange is what leveling added. Read the right-hand column, not the bid form.

Delta submitted the lowest number at $1,142,000. Levelled, they land fourth. The ranking changed, which is the only reason any of this is worth doing.

Five real quotes in five formats, read by the same extractor as anything you import.

The bid set prices its own gaps.

The interesting move is not "read a PDF". It is this: when Sub A excludes crane hoisting and Sub B carries it as a $15,000 alternate, Sub A's gap is worth $15,000. That is a real market price — for this project, this schedule, this site, from a contractor who wants the work. Not a number from a book. Not a percentage.

So the leveling pass looks for competitor evidence first and only falls back to the bundled rate table for your market for elements nobody priced. Every adjustment shows which it used, because an estimator who cannot say where a number came from cannot use it in a negotiation.

Competing bidder. Another bid in this package quoted this work. The strongest evidence there is — a live price for exactly this job.

Their own alternate. They told you what they want for it. Hard to argue with in the negotiation, because they wrote it.

Your award history. What you paid last time, from the award corpus this program builds inside your own workspace as you record what you bought.

Local rate table. 123 subcontract-level rates, revision 2026.1 — an estimate, bundled, overridable, and labelled as such on screen. Used only when nobody in the package priced the item.

Percent of trade. The weakest basis in the product, and it says so. Replace it before you argue it.

Estimator's figure. You typed it. It is marked as yours and it survives every re-run — the engine never quietly overwrites a call you made.

Two rules keep the matrix honest, and both are enforced by tests

Silence on core trade scope is not a gap. A sub quoting "the HVAC work" for a lump sum is carrying the ductwork whether or not their letter enumerates it. Silence only counts where it means something — the universal items like crane and cleanup, and anything a competing bidder took a position on.

Hidden cost is the cost of the award itself. Summing every bidder's adjustments produces a figure several times larger than any cost anyone will incur. You award one subcontract, so the headline figure is the exposure on the bid you would actually award.

Nine ways a bid set fails to be comparable.

Every finding names the bidders on each side, quotes the line it came from, states what it is worth and on what basis, and says what to ask. The counts below are what the demo package actually produced from five PDFs.

One bidder carries the work; another excluded it or never mentioned it. Both get levelled — flagged differently, because an exclusion is on the record and silence is not.

Scope every bidder left out. It doesn't change who is cheapest; it changes what the trade costs you, and it is yours by default.

"If required", "one mobilization", "continuous access", "subject to change" — quoted back at you with the reason each one matters.

HST in or out, bonding, escalation clauses, overtime assumptions, and a validity window that expires before you can award.

A bid priced against a superseded document is a bid for a different project.

A placeholder wearing a price tag. It converts to a real number after award, when you have no competitive tension left.

A bid still well below the field after leveling. In this market that is rarely a bargain.

Priced against a different quantity than the one in the tender documents — the same rate against a different job.

A rate is present but the extension is missing or wrong. Quiet, and it survives a read-through.

40 findings, 14 of them high severity, across 36 scope elements and 5 bidders — 26 rows where the bidders disagree and 10 carried by nobody. $226,160 of adjustment sits on the bid you would award; $84,986 is what awarding on the bid form alone would have cost.

What it does, screen by screen

The figures in this section are the shipping application working the demo package that ships with it: 240 Front Street West, an office fit-out and base building upgrade in Toronto — the mechanical trade, a $1.32M carry, five bidders, five deliberately different formats. The quotes are a fixture, deliberately messy; the extractor and the leveling engine are the ones your own quotes go through.

One number: the cost of not levelling.

The headline is what awarding on the bid form alone would have cost you. Under it, the bid-versus-levelled bar for every bidder, ranked, with the position each one moved. Then a written review that says who to award to and what is still open against them.

  • "Where the hidden money is" totals each scope element across every live bid — the field's exposure across every live bid
  • Levelled award against your carry, with the variance stated as a percentage you can take to a meeting
  • Ask it who should I award to? or what did nobody price? — parsed by rule, previewed before it changes anything

Scope down the side, bidders across the top.

The day in Excel you were about to spend. Every cell says what that bidder did with that scope element — IN, EX, —, ALT, ALLOW, OTH or COND, with the adjustment underneath it.

  • Filter to only the rows where the bidders disagree, which is where the argument is
  • Click any cell to change that bidder's position or override the adjustment — a call you make by hand is marked and survives every re-run
  • Exports to Excel as five sheets with live SUM formulas, so your overrides move the totals in the file too

A quote letter is not a data format. It becomes one.

Drag the quotes onto the window — PDF, Word, Excel, a saved email or a scan. Each is read and split into base price, inclusions, exclusions, qualifications, alternates, allowances, unit rates and commercial terms, with the page it came from and an extraction confidence on every line.

  • Addenda acknowledged is tracked per bidder — 3 of 3, or the gap that is about to cost you
  • Scanned quotes are recognised on the machine and flagged for checking
  • A bidder who never sent anything can be entered by hand; every extracted line is editable

Decide what every bid is compared against.

The checklist is the basis of the whole exercise, so it is visible and editable on the page. 15 trade packages and 18 universal items ship with the app, seeded by trade and then extended from the bids themselves — if a bidder priced something the checklist didn't know about, it gets added.

  • Switch an element out of scope and it stops generating findings — for work genuinely not in this package, not to quieten a finding you would rather not deal with
  • Every element carries its CSI section, whether it is critical, its typical share of trade value, and the rate it is valued from
  • Add your own elements — the ones your projects always argue about, and they persist

The answer, the arithmetic behind it, and the open items.

A recommendation with a status — award, award subject to clarification, or don't award yet — the levelled figure against your carry, and the high-severity items that are still open against the recommended bidder. Four documents come out of this screen.

  • Award recommendation (PDF) — written to be read by a PM who was not in the meeting
  • Clarification letters — one per bidder, every open item with the reason it is being asked. A vague letter gets a vague answer
  • Subcontract scope paragraphs — written from the levelled position, so what leveling added back cannot be quietly re-excluded after signing
  • Handoffs — JSON for the AI Estimator, CSV for ScopeGuard, a Procore-shaped bid tab, and raw JSON for your own tools

Every finding is quoted in full

The words you negotiate against are the words they wrote. Each finding names who carried the work and who didn't, quotes the evidence line from the document it came from, states the value and its basis, and gives you the question to put. Mark any of them real or noise; the verdict persists.

It grades its own work

Record what you awarded and what it finally cost, and the app reports how well the levelled figure predicted the outcome against how well the bid form did. That number is the honest measure of whether this tool is earning its place — computed from your own history, inside your own workspace, and it is allowed to be unflattering.

The failure mode is not a crash. It is a plausible matrix that is wrong.

A levelled number an estimator cannot defend is worse than no number at all, because it is the one they will act on. Four things keep this honest.

Nothing is invented

Adjustments come from the bids first and the rate table second, and every one shows which. MaraponeAI explains, prioritises and drafts — it never decides a number, and it never finds a gap.

Every figure states its basis

Competitor price, their own alternate, your award history, the rate table, a percentage of trade value, or your own figure — ranked from strongest to weakest and printed on the adjustment. You can check the arithmetic in front of the sub.

Silence and exclusion are not the same thing

Both get levelled, because both cost money. They are flagged differently, because an exclusion is on the record and silence is what reaches site unresolved, and because silence on core trade scope is not a gap at all.

You overrule the engine

Change any coverage call, override any adjustment, edit any extracted line, take any element out of scope. Your calls are marked as yours and survive every re-run — the engine reads; you rule.

The model never decides a number.

A tool that hands you a levelled award has to be right about it, because the first invented adjustment you take to a subcontractor is the last time anyone in your office opens the app. So the leveling engine is deterministic: coverage classification, gap detection, valuation, commercial normalisation and severity all run on rules, the other bids, and your own data.

MaraponeAI explains, prioritises and drafts. Every figure it quotes traces back to a line in a document or a row in the rate table. Type set crane to 18,000 on Northgate and the instruction is parsed by rule and previewed before it changes anything.

The engine underneath is deterministic: every adjustment traces to a quote, a line and a basis you can check. Our own model sits on top of that, adding the free-form narrative, and it never gets to decide what was found.

Your workspace, and only yours

Your subcontractors' pricing is the most commercially sensitive data in your office, so it is read in a workspace scoped to your account by our own model, no third-party AI vendor receives it, no other firm's work is pooled with it, and nothing is trained on it.

Fully functional without a model

Every finding, adjustment and recommendation is available from the deterministic engine alone. Without a model you lose free-form narrative — you do not lose a single finding or a single dollar figure.

Every run keeps its own snapshot

Packages, quotes, overrides and your award corpus stay in your workspace, and each leveling run is versioned, so you can open the matrix exactly as it stood on the day you awarded. All of it exports to Excel. Cancel the plan and the exports you have taken stay readable.

You can see every rate it used

Trade scope checklists, the quote-language patterns and your market’s rate table are plain JSON data packs you can open. Every adjustment on a levelled bid prints the line it came from, and the levelling matrix exports to Excel with that basis attached, so a number you cannot defend never leaves the program.

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.

On one mechanical package, awarding on the bid form alone would have cost $84,986, because the low number was not the low cost. That is one trade, on one job.

The Pro plan is $499 a month and runs on every package after that one. That is the whole value argument: you are not buying software hours, you are buying back the day in Excel and the difference between the cheapest number and the cheapest bid.

  • The complete program the moment you sign in — nothing to install, no licence key, no machine to provision
  • Unlimited packages, bidders, quotes and machines in your company. No per-seat or per-project fee, ever
  • 15 trade packages, 155 scope elements, 9 finding types and the 2026 leveling rate table for your market
  • Excel matrix with live formulas, award recommendation PDF, clarification letters and subcontract scope paragraphs
  • Every improvement the day it ships, no version to buy, no upgrade to schedule
  • Any modern browser. Unlocks this tool on the mobile app, Q4 2026, included
  • Cancel any month from your dashboard, no long-term contract — see the plans
  • Your historical awards loaded in bulk, so the engine starts with what you have actually paid
  • A rate table rebuilt for a market other than the one you bought, or checklists for trades you self-perform
  • Wiring into Procore, Sage, Bluebeam or your own bid-tab database
  • Workflows the product doesn’t cover — bid solicitation and purchase orders (prequalification is TradeVet, on the same plan)
  • Bespoke development beyond the eight programs — that is quoted as project work

Buying the product first doesn’t lock you out of a customised setup later — most teams start here, then commission the parts that are specific to them. See all three levels →

Pro opens all eight programs for $499/mo — the five pre-award programs plus FloatGuard, TradeVet and CostLine — with a setup call that runs a real job of yours from the drawing set to the awarded subcontract.