Quoting in the Age of AI, Tariffs and Market Shifts

Paperless Parts co-founder and CEO Jason Ray on stage at the company’s POWER users conference, which took place April 30-May 1, 2025, at the Boston Convention and Exhibition Center. Photos provided by Paperless Parts. 

The topic of tariffs is bound to come up at any manufacturing event right now, and this is especially true at a gathering of those responsible for quoting and estimating. Paperless Parts’ annual POWER users conference, which took place April 30-May 1, 2025, at the Boston Convention and Exhibition Center, was one such event. 

The conference, now in its third year, provides a way for Paperless Parts to update its users on the latest releases, which are pushed out throughout the year. “We were finding that it was incredibly difficult for us as a team to keep our customers up to speed with what we're doing,” says co-founder and CEO Jason Ray. “We felt like everybody was driving it in first gear. So, we started this user conference to get people to drive the car in fifth gear.”

Featured Content

Automation is for Everyone READ MORE

Why Add TASC to Your List? READ MORE

Register for TASC: The Automated Shop Conference READ MORE

Between breakout sessions giving users hands-on experience with the platform and keynotes that introduced new features and gave users a sneak peak of Paperless Parts’ roadmap, panel discussions highlighted software features that are helping users address the challenges manufacturers face today. One existing feature spotlighted during a panel discussion called “Navigating Today's Metal Markets” was the platform’s real-time materials pricing. Paperless Parts works with metals suppliers to integrate material prices, so users can have the most up-to-date information for quotes. “We built this feature in response to the metal pricing fluctuations we saw during COVID,” Ray says. “We think this will be really no different.” The company also has plans to enhance this feature to automatically update users that have sent quotes when price or lead times for the quoted material change. “The ability to do that is kind of like a heads-up display in your car. It's right in front of you. It's a lot safer. It's a lot less risk.”

Not only does the event keep users up to date with the software’s latest releases, but it also serves as a way for the company to solicit suggestions for new features.

The conference also provides a valuable opportunity for Paperless Parts to get feedback from its users on features they’d like to see. “Our customers are builders,” he notes. “They like making things. They're constantly thinking, ‘How can we make this better? How can make it stronger?’”

Dropping a BOM Builder

User feedback provided the basis for one of the new features highlighted at POWER 2025 — the BOM Builder. Ray says that, even before tariffs were implemented on imported materials, he saw margins getting thinner for materials suppliers, leading to them offering more value-added services to manufacturers, such as tube laser cutting, waterjet cutting and other forms of preparing blanks. “You're starting to see more and more of the material providers provide the easier processes,” he explains. “If I can take a bar and cut that bar into chunks, instead of you running it on your saw, you should take that.” This frees manufacturers from having to buy and maintain equipment such as saws, EDMs and waterjet cutters which they might not use that often. However, this change is causing manufacturers to shift their capabilities downstream into assembly.

At the same time, Ray sees OEMs trying to consolidate supply chains and asking manufacturers to provide assemblies instead of just components. “When you talk to supply chain managers, the main thing you're hearing is that supply chain complexity leads to a lack of resiliency. So, they want to work with a one-stop shop,” he notes. “They want to be able to go to you, and they want you to be able to produce the entire thing.”

Paperless Parts product manager Meg van Deventer presented on the company’s new BOM Builder feature, which is designed to make building assemblies within the software faster and more efficient.

Paperless Parts users already had the ability to build out bills of material from 3D models, but Ray acknowledges that the process was difficult when working from PDFs. “We sat the team down and said, ‘Go build this bill of materials from this print and count the clicks, understand the problem, and come back and let's talk about it,’” he says. “And everybody came back unanimously and said, ‘We have to fix this.’”

Adding this feature wasn’t easy. “This is probably one of the single biggest investments we've made in our product, because it relied on several different underlying technologies that we needed to build out,” Ray says. The company had to figure out how to enable the software to read prints, identify tables, extract the tables from the prints and then move the data from the tables into other areas of the user interface. It developed optical character recognition (OCR) vision and classification capabilities in house, which kept costs down and enabled the product to maintain FedRAMP (Federal Risk and Authorization Management Program) compliance. Then the company worked with 25 of its users to develop and fine-tune the BOM Builder before its release in April.

Aided By AI

The BOM Builder uses AI to help extract the tables from prints and spreadsheets. From Ray’s perspective, for now, AI is best used in cases such as this to help inform decisions, not to make decisions. So instant quoting isn’t on the company’s roadmap, but he notes that some customers have successfully built instant quoting on top of Paperless Parts using APIs. “I think it works for a very small subset of work, but I think it creates more risk today than value,” he says.

Paperless Parts has been very selective about where and how it uses AI. Ray wants Paperless Parts users to not only trust that the output of the AI is correct, but also that Paperless Parts is going to be responsible with the data they’re providing. “There’s a lot of trust to be had there,” he says.

So, for now, Paperless Parts is only using AI for tasks that have an objective answer, and there’s plenty of objective answers to be found in quoting. “It's like tapped holes,” Ray says. “How many tapped holes are there on the part? That cannot be a subjective, guessed answer. We've identified these specific features of the part, and that is what is driving the cost. So that's where we're starting with AI. We're not going to do any subjective, derived responses to our customers.”

The next step is to use data to gather insights into jobs. For example, users will be able to see that margins drop when certain industries request certain numbers of parts or certain tolerances. “That's where I would like to go, because that is tribal knowledge that estimators have a gut feel for,” Ray explains. This is the type of information that’s difficult for estimators to pass down because it has historically been difficult to articulate the reasons behind it. “Our goal is to start to derive those same sorts of outputs that are rooted in data, so it’s objective, and then you can trace it. We saw this on a print, and we lost that job, or we didn't make money on that job.”

According to Ray, the BOM Builder represents a significant investment from the company because it required the development of multiple underlying technologies, including the ability to read prints, identify tables and extract them from the prints. This technology serves as the basis for its next technology release, a GD&T call-out extraction feature.

Extracting More Data

The capabilities Paperless Parts developed for the BOM Builder laid the foundation for its next release, a GD&T call-out extraction feature. It started with pulling in part numbers, revisions, descriptions, materials and specifications. “Then we said, ‘Well, what about symbols?’ It turns out that most CAD systems have a couple hundred fonts,” Ray explains.

This poses a difficulty for automated data extraction. Fortunately, Paperless Parts was well-positioned to take on the challenge. “We have samples of drawings from every single CAD system in almost every single font,” Ray notes. Developers used this to build synthetic training data that enabled them to train AI to identify this data accurately.

According to Ray, this data will drive automation in the quoting process. For example, if the drawing calls out brazing or honing, Paperless Parts users can now set up a rule where that quote now automatically adds an operation or kicks off an RFQ to an outside supplier if that process can’t be done in house. If the drawing calls out parallelism, users can set up a rule to automatically change the type of material they need to buy. Or if the datum structures go past E, users can set up a rule to automatically decrease yield by two percentage points. “You can start to build these rules where the user doesn't need to know how to write code in order to configure the system and put their tribal knowledge into Paperless Parts,” he explains.