Somewhere around demo number three, one of our guest judges said: "We're only two demos in and I’m inspired."

By the end of ten presentations, that same judge had goosebumps.

For four days, the Ediphi team stepped away from the roadmap, formed teams around ideas they're excited about, and built something new; no matter how ambitious, unpolished, or out-there it might be. 

Then everyone got on a call and showed their work to the most seasoned panel of preconstruction professionals we could find.

What is Ediphi’s Make-a-thon?

Every year, we drop what we’re doing to build for a week; we call it a “Make-a-thon”. 

It gives our people the chance to experiment, build, and collaborate on something brand new for the product and our customers. The only rule is that it has to be Ediphi-related; beyond that, anything goes.

We do it for three reasons:

  1. Cross-team collaboration and culture. As a remote team, Make-a-thon is one of the few times everyone, from any role, builds together, side-by-side.
  2. Leaps in innovation. Could something that normally takes three weeks get built in three days?This year, many projects showed up close to shippable.
  3. Product value. Every idea has to answer one question: how does this make Ediphi better for the people using it?

This year raised the stakes on the judging panel, too. 

Joining Ediphi's own Sarah DeBaca and Mike Navarro, two guest judges came straight from the field: Sarah Forrest, Chief Estimator and Preconstruction Director at Hensel Phelps, and Taylor Borders, Regional Preconstruction Director at AECOM Hunt.

We may be biased, but these really are two of the sharpest minds in preconstruction and they pressure-tested every idea against what actually happens on a jobsite.

Here's what our teams showed them.

1. An AI assistant for support tickets

Meet “Sherlock”. This AI-bot sits inside Help Scout and automatically reads new tickets, diagnoses what the customer needs, and gives a suggested resolution.

In the demo, it found a 74% match to previous tickets and closed the case in a fraction of the usual time.

"He's here as an assistant, as a resource," the team explained. 

Sherlock doesn't answer tickets itself — a human still reviews everything before it goes out. 

Why it matters: Ediphi already responds to tickets in about ten minutes on average. Sherlock's goal is five, without removing the human from the loop.

2. Live, secure links for anyone outside Ediphi

Estimating involves plenty of people who'll never log into Ediphi (yet!). Which means for Ediphi users today, you need to export a static PDF to share a project. 

“Edishare” replaces that with one button: a secure, read-only link to a live cost model, cost trend line, or variance report — no login required, reusable or one-time-use, revocable anytime.

The demo doubled as one of the day's best bits — a fully produced Inception-style short film, with live share links dropped into the judges' chat mid-presentation. "This is legit," one judge said. "Pretty slick, ain't it?"

Why it matters: You don’t need to re-exporting a file every time a number changes. Anyone outside Ediphi gets a live link into it instead.

3. Native, live dashboards inside Ediphi

Estimators already build cost dashboards in Excel.

One team brought that experience natively into Ediphi: spin up a dashboard from a template or from scratch, with full cross-filtering between charts, then duplicate it, save it as a company-wide template, or share it with view-only guest collaborators.

The moment that stole the room: a fully published support documentation article, ready before the demo even ended. 

"You built the support article already?" one teammate asked. 

Sarah Forrest put it simply: her team would want this in production immediately.

Why it matters: You don’t need to rebuild the same Excel reporting templates for every project. The full cost story lives and updates right where the estimate does.

4. Real-time comments, built into the estimate

The answer to “Why did this line-item jump 18%?" happens across Slack, Teams, or email. But does it have to?

One team built conversations into the estimate!

Right-click any cell to comment and tag a teammate and start a thread. A notification center tracks every comment across every estimate and clicking one jumps straight to the line it's attached to.

"Was that Figma, or actually in-app?" a judge asked, mid-demo. "That's real life," the team answered. 

“It's like you read our minds,” said Sarah.

Why it matters: Conversation and collaboration happens alongside the number it's about.

5. AI-powered bid leveling

Subcontractor bids arrive as PDFs and every proposal is a different format. This team's fix takes any proposal and pulls it into a clean, cited, reviewable scope list automatically, with every parsed line linking back to its source in the PDF.

To keep things honest, the team asked judges Taylor Borders and Sarah Forrest to send real, messy subcontractor quotes to parse live on the spot. 

Taylor's feedback was the sharpest of the day: it's these types of messy mechanical and electrical bids where a GC's exposure lives. Saving time and building assurance here is directly correlated to profit.

It's what earned them the day's most memorable reaction: "We have the best team in the world," one Ediphi leader said. "I have goosebumps on my arms right now."

Why it matters: Ediphi's estimates are already structured by scope in a way nothing else in the market is — this turns an unstructured PDF into that same structure in minutes.

6. Buy materials without leaving your estimate

One team made buying materials fun — and AI-powered. 

They ventured to turn a finished estimate into “a shopping list”, grouped by bid package, ready to buy from suppliers with live pricing and stock on hand.

For their demo, the team built a real RFQ, sent it to three suppliers, and parsed an actual returned quote PDF, verifying it against the source and leveling it against the estimate — green where you beat your number, red where you didn't. Award the winner, and the purchase order drafts itself, with actuals flowing back onto the estimate without touching the original number. 

Asked point-blank whether the PDF-reading AI was real or faked, the team didn't hedge: "It was real."

Why it matters: Real-time variance between what you estimated and what you actually paid, in the same grid the estimate lives in.

A few more worth a mention

7. Use Group Therapy turns a single owner RFP PDF into a defensible cost model in minutes. It builds from individual use groups instead of forcing one project's benchmarks to stand in for all of them, with AI suggesting the best-matching historical data along the way.

8. Millisecond Removers recreated the estimate grid to make edits roughly 7x faster (and it’s already sitting at about 80% production-ready!)

9. Operation Seal fixed a rounding bug where large quantities got rounded down instead of up (think: 1.35 toilets rounding to one instead of two), which can mean millions in under-procurement on big jobs. Rounding is now its own configurable setting that flows automatically from the Unit Price Catalog to every new line item.

10. La Roja gives every pull request its own live, clickable preview link automatically; tag the PR, and a couple minutes later you've got a working version of that exact branch to click through with no setup required.

What’s next

Ten ideas, four days, and a panel of judges who see mediocre estimating software for a living and don't get impressed easily.

They were anyway. By the judges' own account, "Every team came up when we were discussing.” 

Not every project that comes out of Make-a-thon ships, but some of these are very close…

Some dashboards shipped with a support article ready to go before the demo even ended, and several other teams are already talking through what it would take to get their work in front of customers.

Which is the real point of the week: give a customer-obsessed team space to build, throw in some genuine constraints, and hand the results to some of the sharpest minds in preconstruction — and you get software that's already been pressure-tested by the people who'll actually use it.