A Full Electrical Takeoff From a Phone on a Saturday Night

How an estimator ran a complete electrical takeoff from his phone on a Saturday night, with an AI assistant named Gandalf

Saturday Night, Phone in Hand

Picture this: Saturday night. Family on the couch. Phone in hand.

That's where Jon was when he decided to run a full electrical takeoff on a 24-page DD set.

No laptop. No Bluebeam. No counting symbols at a desk for hours. Just a phone, a Slack channel, and an AI assistant the team calls Gandalf.

Meet Gandalf

Before we get into the numbers, a quick detour on the name.

Gandalf is Boon's AI estimating assistant. The team named it after the Tolkien wizard for reasons that felt obvious at the time: it shows up precisely when needed, it knows things no one else thought to check, and it occasionally performs what can only be described as magic with construction drawings.

The name stuck. And like its namesake, Gandalf has a habit of surprising people.

Jon, an electrical estimator with decades of experience, has become one of Gandalf's most active users. He's described himself as "obsessed" with it. Not because it replaces his expertise, but because it handles the parts of the job that eat his time without engaging his brain.

So on this particular Saturday night, Jon opened Slack on his phone and said: "Let's go."

The Session

Over the next four hours — while watching TV, hanging out with his family, and checking in every 20-30 minutes — Jon directed Gandalf through a complete electrical takeoff.

The workflow looked like this:

Jon uploaded a 24-page DD electrical and fire alarm set
Gandalf read every page — general notes, legends, schedules, riser diagram, fixture schedule

Jon gave high-level direction: "Start with power plans." "Now do fire alarm." "Add RFIs to the Excel."
Gandalf executed: counting, extracting, cross-referencing, formatting

Jon made the strategic calls. Gandalf did the counting.

The Results

From a single 24-page drawing set, Gandalf produced:

A full high-level project review with 7 suggested RFIs and 20 bid qualifications flagged before a single number was crunched
517 receptacles detected and counted across all power plans
~276 fire alarm devices broken down by floor
50 lighting fixture types extracted from the fixture schedule
An 8-tab Excel BOM with live formulas covering receptacles, branch wiring, home runs, panel schedules, fire alarm devices, and feeder schedules

All from a phone. All on a Saturday night.

When the AI Hit a Wall, It Kept Going

Midway through the session, Gandalf ran into an authentication issue that would have killed most automated workflows. The AI takeoff engine went down for about two hours.

Instead of stopping and waiting for help, Gandalf automatically switched to manual image analysis — scanning each floor plan with vision AI and continuing the count. Jon didn't have to troubleshoot. He didn't even notice until he read the session log afterward.

That's the difference between a tool that assists and a tool that adapts. A script would have stopped. Gandalf found another way.

What This Actually Means for Estimators

Let's be clear about what's happening here, and what isn't.

No one is saying estimators are getting replaced. The judgment calls, the experience, the "something looks off about this drawing" instinct — that's irreplaceable. That's what decades of field and office time builds.

What IS going away is the grunt work. The hours spent counting symbols. The tedious cross-referencing between schedules and plans. The manual data entry into spreadsheets that should have been automated years ago.

Jon stayed strategic the entire session. He decided what to run, flagged what to include, and made the calls that required human expertise. Gandalf handled counting, extraction, and formatting.

As Jon put it during a recent team call: "I'm giddy every time I use this thing."

That's not the reaction of someone worried about being replaced. That's someone who just got 20 hours of their week back.

The Numbers

Consider what a traditional version of this takeoff looks like:

Manual approach: 2-3 full work days at a desk, counting symbols, cross-referencing schedules, building spreadsheets from scratch
With Gandalf: 4 hours of intermittent check-ins from a couch

That's not a marginal improvement. That's a fundamentally different way of working.

And the output wasn't rough or approximate. It was a structured, formula-driven Excel workbook ready for pricing. The kind of deliverable that used to require a full day of formatting alone.

"There's People That Are Just Excited About Gandalf"

Jon said something on a recent call that stuck with us: "If you meet Dave's expectations, you're exceeding others... There's people that are just excited about just Gandalf. They don't even care about the rest."

That tracks with what we're hearing across the industry. Estimators aren't looking for a tool that does everything. They're looking for a tool that does the repetitive parts well enough that they can focus on the work that actually matters: identifying scope gaps, catching design conflicts, making informed bid decisions.

The construction industry has a well-documented problem: 32% of cost overruns trace back to estimating errors, and estimators spend 20+ hours per week on repetitive tasks that could be automated.

Jon's Saturday night takeoff isn't a gimmick or a demo. It's what the work actually looks like when you remove the bottleneck.

What Estimating Looks Like in 2026

An estimator kicked off a full DD review and initial BOM from his phone on a Saturday night. No laptop, no Bluebeam, no counting symbols by hand. He gave high-level direction and got back a bid-ready workbook.

The AI counted 517 receptacles, extracted 50 fixture types, built an 8-tab BOM with live formulas, and recovered from a technical issue without human intervention.

Honestly? It felt pretty good doing it from the couch.

Boon AI automates construction takeoffs, reducing takeoff time by up to 90%. Learn more at getboon.ai

Embedded Videos: