Tool checks 2D drawings and 3D models automatically, cutting errors
17 July 2025 
A Canadian software developer has released an AI tool that reviews 2D drawings and 3D models automatically, thus ensuring quality, boosting productivity, and reducing errors and costs. CoLab says that its AutoReview tool can automate repetitive design checks, allowing engineers to focus on value-added design work.
Currently, engineering teams currently spend thousands of hours every year reviewing drawings. The new tool uses AI to automate common checks, thus saving time.
CoLab surveyed 250 engineering leaders and found that they spend an average of 23% of their time on non-value-added work. “Even if AutoReview only replaces that portion, teams could reclaim nearly a quarter of their time,” points out the company’s co-founder and CEO, Adam Keating.
AutoReview works by scanning documents and identifying common issues, such as errors in drawings and DfM (design for manufacturing) optimisations. For example, it will flag if there are missing or incorrect countersinks in drawings. It can also check moulding measurements to ensure that wall thickness dimensions will work and that moulds are easy to remove for manufacturing.
The software can also learn from earlier designs using similar geometries or parts, helping to avoid costly mistakes being repeated in different designs. CoLab says this can deliver significant savings by reducing scrap rates and warranty claims.
“If a sharp internal corner previously caused fatigue failures and warranty claims, AutoReview can flag similar geometry in a new design and surface that past feedback for the current design team to use,” Keating explains. “As product complexity increases, it’s easy for these types of errors to slip through the cracks. But ultimately these are preventable costs.”
In complex manufacturing sectors such as aerospace, missing one dimensional check can mean reworking a drawing or delaying critical milestones. CoLab says the new tool can replace redundant manual checks and deliver greater consistency and adherence to company-specific design rules, validating 2D drawings and 3D models in programs such as SolidWorks, Catia, Creo and NX.
“Validators in design teams currently spend hours checking every surface for proper dimensions, but AutoReview can perform these checks within minutes,” says Keating. “This gives engineers more time to focus on value-add work, like bills of materials (BOM) cost reduction – savings that are critical to help offset the tariff and supply chain pressures these companies are exposed to.”
Beyond adherence to standards, the tool can also identify DfM optimisations automatically, helping to reduce scrap, prototyping and tooling costs.
Beta versions of the software are already being used by global manufacturers including Techtronic Industries (the $14bn parent company of Ryobi), and the wind turbine blade manufacturer, TPI Composites. Both companies are testing its abilities to catch design issues earlier.
“AI won’t replace engineering teams. But it will take on the repetitive decisions that slow engineers down, so they can spend more time solving hard problems and pushing the product forward,” Keating concludes. “It will also protect engineering organisations from the wave of retirements and the threat of losing institutional knowledge.
“Fast-forward a few years, and engineering companies that don’t leverage AI to automate administrative tasks, prevent errors and reduce costs, won’t stand a chance.”
The new tool is based on CoLab’s Design Engagement System (DES) which helps engineers to run faster, higher-quality design reviews, while capturing organisational knowledge as a by-product of their day-to-day work. DES is already being used by customers including Schaeffler, Johnson Controls and Mainspring Energy, which have cut their BOM costs by up to 50% and accelerated their design cycles by 100%, allowing them to launch products months sooner than normal.
AutoReview runs inside CoLab’s secure environment and works with existing PLM infrastructure, ensuring that design IP is protected.
