AI for Engineering Knowledge Management
Younger engineers struggle with GD&T and tolerancing. Learn why the knowledge gap exists, the costly mistakes it causes, and how AI is closing it.
·
⏱
8 min read
Michelle Ben-David
Michelle Ben-David is a mechanical engineer and Technion graduate. She served in an IDF elite technology and intelligence unit, where she developed multidisciplinary systems integrating mechanics, electronics, and advanced algorithms. Her engineering background spans robotics, medical devices, and automotive systems.
BOTTOM LINE
The GD&T knowledge gap isn't going away on its own. Universities can't teach 20 years of judgment in a semester, and the engineers who carry that judgment are retiring faster than companies can replace them. The practical answer isn't more training alone. It's giving every engineer instant access to the tolerancing decisions, standards, and design history that already exist inside your organization. Leo AI turns that scattered knowledge into something every engineer on the team can tap into, whether they've been there 20 years or 20 days.
There's a quiet crisis on the shop floor that nobody talks about in quarterly reviews. A junior engineer submits a drawing. The tolerances look reasonable. The part gets manufactured, shipped, and assembled. And then it doesn't fit.
Not because the design was bad. Because the GD&T callouts were wrong.
Geometric Dimensioning and Tolerancing is one of the most critical skills a mechanical engineer can have, and it's also one of the least well-taught. University programs rush through it in a week or two. On-the-job training assumes someone senior will catch mistakes. And increasingly, that "someone senior" is retiring, switching companies, or too overloaded to review every drawing that crosses their desk.
The result? Tolerancing errors that don't show up until parts hit inspection or assembly. Scrap, rework, vendor disputes, and delays that cost real money. And a growing gap between what junior engineers know about GD&T and what they actually need to know.
The Tolerancing Skills Gap Is Wider Than You Think
GD&T is supposed to be foundational. Every mechanical engineering program covers it. But "covers it" and "teaches it well" are two very different things.
Most undergraduate programs dedicate maybe 10 to 15 hours to GD&T as part of a broader engineering graphics or manufacturing course. Students learn the basic symbols, maybe work through a few textbook exercises, and move on. They graduate knowing that position tolerance exists, but they've never had to choose datum references for a complex multi-feature part.
The real education happens on the job. A senior engineer reviews your drawing, circles the callout that doesn't make sense, and explains why your datum scheme won't hold up during inspection. You learn by making mistakes and having someone catch them. That system worked for decades because the ratio of experienced engineers to junior ones was manageable. That ratio is shifting fast.
According to industry estimates, roughly 25% of the manufacturing engineering workforce will reach retirement age by 2030. The engineers who carry the deepest GD&T expertise, the ones who've spent 20 or 30 years interpreting drawings and debugging tolerance issues on the floor, are the same ones walking out the door. And they're taking decades of practical tolerancing knowledge with them.
IN PRACTICE
The technical Q&A feature pulls from real engineering standards with source citations, giving engineers confidence they're getting accurate, relevant answers. It has noticeably cut weekly meeting load and reduced the number of 'can you find X spec' emails significantly.
— Verified User, Mechanical or Industrial Engineering, Small Business
The Five GD&T Mistakes That Cost Companies the Most
Not all tolerancing errors are created equal. Some are cosmetic, some are embarrassing, and some are genuinely expensive. Here are the five that keep showing up across industries.
Over-tolerancing. This is the most common mistake junior engineers make. They specify tight tolerances everywhere because tighter feels safer. But every unnecessary tenth of a thou drives up machining cost, slows production, and increases scrap rate.
Wrong datum selection. The entire GD&T framework depends on choosing the right datums, the reference features everything else is measured from. Pick the wrong primary datum and every callout downstream is measuring from the wrong baseline.
Misapplied material condition modifiers. MMC, LMC, and RFS change how much positional tolerance a feature gets based on its actual size. Junior engineers often default to RFS because it's the simplest to understand, leaving performance and cost on the table.
Conflicting or redundant callouts. When a part has both dimensional tolerances and geometric tolerances controlling the same feature, the tighter one wins. Engineers who don't fully understand the interaction end up with drawings that are internally contradictory.
Ignoring the inspection method. A tolerance is only as good as your ability to verify it. Engineers sometimes specify GD&T callouts that are theoretically correct but practically impossible to inspect with available equipment.
Why the Training Pipeline Is Broken
If the mistakes are this predictable, why do they keep happening? The training pipeline has three fundamental breaks.
Break one: Universities teach syntax, not judgment. GD&T courses teach you what the symbols mean, not when to use them. Students learn that a flatness callout controls form deviation, but they don't learn how to decide whether flatness, parallelism, or profile is the right choice for a given sealing surface.
Break two: ASME Y14.5 is dense and evolving. The governing standard for GD&T in the US is ASME Y14.5. It's over 300 pages of precise technical language. The 2018 revision introduced significant changes. Many engineers learned from the 2009 version and never formally updated.
Break three: Mentorship doesn't scale. The traditional path to GD&T proficiency is apprenticeship. You sit next to someone who knows what they're doing, and you learn by osmosis over several years. That works fine when you have two junior engineers and five senior ones. It falls apart when you have ten junior engineers and one senior engineer who's also handling design reviews, ECOs, vendor issues, and their own project deadlines.
The Real Cost of Getting Tolerancing Wrong
The financial impact of GD&T errors is hard to isolate because it shows up in so many places. It's not a single line item on a budget report. It's scattered across scrap costs, rework hours, vendor negotiations, delayed shipments, and engineering change orders.
Consider a typical mid-size manufacturer producing precision assemblies. A single tolerancing error on a critical mating feature can cascade through the entire assembly. The part gets manufactured to the drawing, passes individual inspection, but fails at assembly because the tolerance stack-up wasn't analyzed properly.
Any of those options costs time and money. Rework on a batch of CNC-machined components can run $5,000 to $50,000 depending on complexity and volume. If the error makes it past incoming inspection and into a shipped product, the cost multiplies by an order of magnitude.
Then there's the hidden cost: the engineering hours spent debugging tolerance issues that should have been caught at the drawing stage. Senior engineers spend a significant portion of their time reviewing and correcting junior engineers' GD&T callouts. That's high-value engineering talent doing quality control instead of advancing new designs.
How AI Is Starting to Close the Gap
The traditional answer to the GD&T knowledge gap has been more training. Send engineers to a three-day ASME Y14.5 seminar. Buy everyone a copy of the GD&T reference book. These all help, but they don't solve the fundamental problem: judgment takes years to build, and you can't fast-track experience.
What's changing now is the availability of AI tools that can act as a persistent, always-available reference layer for engineering teams. Instead of waiting for the senior engineer to be free for a design review, an engineer can query an AI assistant trained on engineering standards, company design guidelines, and historical drawing data.
The key differentiator is context. General-purpose AI tools can tell you what a flatness callout means. Engineering-specific AI, like Leo AI, can do something more useful: it can reference your organization's past drawings, look at how similar features were toleranced on previous successful parts, and surface the relevant ASME standards with full citations.
This doesn't replace the need for experienced engineers. It scales the knowledge that experienced engineers have already embedded in their work. Every drawing a senior engineer has ever reviewed, every tolerance decision captured in a released design, becomes queryable institutional knowledge rather than tribal memory locked in one person's head.
FAQ
ASME, "ASME Y14.5-2018: Dimensioning and Tolerancing," 2018
Deloitte and The Manufacturing Institute, "Creating Pathways for Tomorrow's Workforce Today," 2021
Close the GD&T Gap
Give every engineer access to your best tolerancing decisions.
Leo AI connects to your PDM and design history, so engineers can find how similar features were toleranced on past designs. Fewer mistakes, less rework, faster reviews.
Schedule a Demo →
#1 New AI Software Globally - G2 2026
Enterprise-grade security
Trusted by world-class engineering teams
Close the GD&T Gap
Give every engineer access to your best tolerancing decisions.
Leo AI connects to your PDM and design history, so engineers can find how similar features were toleranced on past designs. Fewer mistakes, less rework, faster reviews.
Schedule a Demo →
#1 New AI Software Globally - G2 2026
Enterprise-grade security
Trusted by world-class engineering teams