Rules raise flags. People decide causes.

Who built this

Costwitness was built by one cost consultant with twenty years on the commercial side of construction contracts. I spent those years doing this work by hand: reviewing pay applications, certifying payment, pricing change orders, closing out contracts, and the monthly argument about what a contractor's application really supports. Contract forms differ from job to job. The arithmetic, and the argument, do not.

That is the whole reason the product is shaped the way it is. Every rule on every screen exists because I was caught by the thing it prevents. The baseline is frozen because I have watched a schedule of values change quietly between months. The two contingency funds are never added together because I have seen what happens when they are. The notice clock exists because a good claim lost on a date is the most expensive kind of loss there is.

I am not engaged as your adviser and I do not advise on your contract. I am not a substitute for the cost consultant, owner's representative or counsel you already use, and I do not hold professional credentials in the United States. What I build is the record. The judgment stays with you and with the people you pay for it.

Why this product exists

An owner on a guaranteed maximum price contract receives a monthly summary written by the party that spends the money. When we reviewed those summaries for owners, the work was always the same: rebuild the position from the documents, line by line, and compare it to what the summary said. The rebuild was the value. The product is that rebuild, kept every month instead of performed once, so the comparison is always available and the history is always stored.

That is also why the product is narrow on purpose. It manages the inside of the GMP contract. It is not accounting, not a payment system, not a replacement for Procore, and not a development budget tool. A narrow tool can hold a hard rule on every screen. A wide one cannot.

How the baseline is set

Every project starts with the executed GMP amendment and its exhibits being read line by line: the schedule of values, the allowance schedule, the qualifications and clarifications list, the shared savings clause, the notice provisions. Each line is entered with a basis flag: priced, allowance, or assumption. The output is the GMP Baseline Report, a document the owner keeps, with the soft lines ranked by exposure.

This is deliberate and it is the part we refuse to automate. A form cannot tell an assumption from a priced line, and the basis flag drives every risk read downstream. The setup is covered by the pilot on the pricing page, and it is why the figures the product produces afterward can be traced back to a document somebody signed.

What the reviewer does not do

The reviewer does not decide cause classifications, does not approve draws, and does not write the owner's position summary. Those are the owner's decisions under the owner's contract. The audit trail records who made them.

The rules the product holds itself to

Every figure traces to a document, not to a summary. Two contingency funds are never shown as one total. The baseline is versioned, never edited in place. Monthly history is stored as snapshots and never recomputed from current data. Every change to a cause classification is written to an audit table with a name and a timestamp. Money is stored in whole dollars, displayed in a tabular face, and never rounded twice.

The charts follow the same discipline. Never a pie, a donut or a gauge. Never a legend where a direct label fits. One accent color per chart, with the flag color reserved for exposure and overrun. Every chart is inline SVG, so a printed report looks exactly like the screen. Every chart carries a caption naming the source of its figures.

No narrative is generated by software. The rules raise flags. The one page of plain language in every monthly report is written by a person, and the report says by whom.

What we will not claim

You will not find a savings percentage, an hours saved figure, or a customer count anywhere on this site. We have not measured those things, so we do not state them. Where an industry pattern is referenced, it is labeled as a recognized industry pattern and not as our measurement. The calculators take your figures and show what they imply. The one comparison we publish is arithmetic you can check yourself.

The same rule applies inside the product. Benchmarks stay off until sourced ranges exist. Fictional projects are labeled fictional. The footer of every report says what the document is: a record and a calculation from information provided, not legal advice.

Where to go from here

Read how the nine modules fit together, check where the data lives and who can read it, or put your own numbers into the free calculators. If you want to see the ledger on a project shaped like yours, request a demo.