Eight weeks of bid review inside the electrical group of an Upper Midwest specialty contractor.
“It’s a checkpoint. It’s another me going through, looking at our estimates and projects and pointing out the things that can be very costly if missed.”
An electrical estimating group ran Makeoff bid review on live competitive bids for eight weeks and confirmed five catches their own review had not surfaced. The estimator who owned each project verified every one before pricing closed. Two of the five were sitting in addenda that contradicted a drawing note the team had already priced against.
The team worked inside their existing process. Estimators exported the takeoff from Accubid, uploaded the plans and specs, and let Makeoff read the set together before the bid went out. The chief estimator called it dump and go.
Verified on live bids during the pilot, each confirmed by the estimator who owned the project.

“Just the site lighting plans, we had a whole bunch of light poles. It was flagged, it was wrong, so we were able to go in there and adjust that.”
Across the pilot the money added up.
“We found some pretty substantial dollar amounts of items that we did not have covered with our initial estimates.”
“It helps you sleep at night. It’s all the things you worry about in the background, that keep you up at night.”
Every one of these was obvious to the estimator the moment they saw it next to its source. The difficulty is structural. Read on its own, each document in a bid package looks finished, and a conflict only shows up when you hold two of them against each other. That comparison takes time, which is the first resource to go in a week with four bids due, and the documents keep changing while it happens.
Addenda make it harder. They arrive late, sometimes the same week as bid day, and estimators read them for the thing they are expecting rather than checking them against everything already priced. Two of the five catches came in that way, including the pole heights.
Everything on this list was stated plainly somewhere in the package. What turned each item into risk was a second document saying something different, in a place nobody had reason to look.
This team came to Makeoff after an internal attempt.
“We were deeply interested in AI, how can AI assist us with what we do. We had started to dabble in it. One of our guys is fairly in tune with working with AI, so he was trying to build something. We were stumbling through processes.”
The gap between a general purpose model and a bid review that reads a plan set the way an estimator does is most of the work. That gap is what took eight weeks of live bids to close, and it is why the checks below could be written by estimators rather than engineers.
Midway through the pilot the chief estimator and a senior estimator started writing checks of their own. Five of them, built in the customizable Issue Flagger, encode how this firm bids its work. Makeoff runs those five on every project the team uploads now, including the ones that arrive in week three of a crunch.

The estimators asked for things while they worked and we built them. Four shipped inside one quarter.
“They’ve taken ideas that we’ve had and very rapidly brought them into fruition.”
Nothing about the pilot required the team to change how they estimate.
“A very simple uploading of the drawings, the specs, you can export your takeoff. Through a couple minute upload time, seeing the quick and easy results.”
“The system itself is very, very user-friendly. Impressive with how fast it works.”
Makeoff is a review layer for electrical estimating. It reads the drawings and the specs together, the way a senior estimator would with unlimited time, while there is still room to price what it finds. The estimator owns every number, and Makeoff catches what falls between the documents on the way to bid day.
It is the pass where the drawings, the specs, the addenda, and the takeoff get held against each other before pricing closes. Takeoff counts what is drawn. Review asks whether the documents agree about what is being bought.
Because each document reads as complete on its own. A conflict only exists across two of them, and finding it means reading both at once against a takeoff that is still moving.
Bid week is normal. Two catches in this pilot arrived that way, and the pole height change would have cost the contractor after award.
This one tried. They had someone comfortable with AI working on it and described the process as stumbling. Reading a plan set the way an estimator reads it is where the effort sits.
No. Every number belongs to the estimator. Makeoff flags what the documents disagree about and hands it back with the source.
Send a project your team has already bid. The full document set plus your takeoff is all we need. We will run it and walk you through what Makeoff flags, including the items you caught first.
Send us a bid you already priced