AI for CAD Tools
Siemens ships AI in NX, but it is narrower than it sounds. What the Design Copilot reads, what Teamcenter Copilot handles, and where neither one reaches.
·
⏱
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
Siemens ships real AI in NX, but it is narrower than the word copilot suggests. The Design Copilot is grounded in Siemens documentation, which makes it a fast and reliable guide to the software itself and nothing more. Select Similar Faces and Edges and PMI prediction do touch geometry, and both are scoped to a single model. Anything about your product data lives on the Teamcenter side, behind a separate product and a separate licence. The material that spans systems, meaning the standards, the legacy vault, the calculation sheets and the supplier documents, sits outside both. Treat the built-in tools as very good at what they are grounded on, and plan separately for the retrieval problem, because that is the one costing your engineers hours every week.
Siemens has shipped AI into NX, and the first question every design engineer asks is the practical one: does it know anything about the model I have open? The honest answer is that some of it does and most of it does not, and the difference is not a detail. Siemens has deliberately split its AI across two products that reach different data, and knowing which one you are talking to decides whether you get a useful answer or a confident description of a menu command. What follows is a walk through what actually ships in NX today, what it reads, where the boundary sits, and what you still need something else for.
What the NX Design Copilot Actually Answers
The Design Copilot in NX is a natural-language assistant built to speed up learning. Siemens describes it as a way for users to find answers to technical queries, best practices and documentation quickly and efficiently, and positions it for newcomers and experienced users alike who want to discover capability they did not know was there.
The scoping language matters more than the framing. Siemens says the copilot converts user input into domain-specific commands backed by factual information drawn directly from its own documentation. That is a precise claim, and it is worth reading twice. The grounding source is Siemens documentation. It is not your part file, not your assembly, and not your project history.
In practice that makes it genuinely good at a specific class of question. How do I create a swept blend with a guide curve. Which command gives me a variable-radius blend. What is the recommended workflow for a synchronous modelling edit on imported geometry. These are questions engineers previously answered by searching a help portal or asking the person two desks over, and a grounded natural-language interface over the documentation improves on both.
What it is not built to answer is anything that depends on your data. Why the wall thickness on this housing is 4 mm, which revision changed the bolt pattern, whether your company allows this fastener, or which existing part already solves the problem. Those answers do not live in Siemens documentation, so an assistant grounded in Siemens documentation cannot reach them.
IN PRACTICE
It integrates directly with PLM and existing workflows, making past designs, standards, and calculations instantly available. The result is fewer errors, faster decision-making, and a more consistent process across teams.
- Sergey G., Board Member
The AI That Works on Geometry Instead of Conversation
The copilot gets the attention, but the AI features that touch your actual geometry are separate, and they are quieter.
Select Similar Faces and Edges uses AI-enabled recognition to find geometry that matches a selection. You pick a face or an edge, adjust a sensitivity control, and NX finds the other features in the design that match. From there you can edit them in bulk, or define them as reusable regions so that repeated changes are applied once. On a part with forty identical fillets, or a plate carrying a repeated hole pattern, this removes a tedious job that engineers have always done by hand.
The scope is the important part. Siemens describes this as operating within a single part file. It recognises geometric similarity inside the model you are in. It does not reach across the assembly, and it does not compare your part against the thousands of parts your company has already released.
The second geometry feature is PMI prediction. Siemens uses domain-specific machine learning models to predict and assign the correct annotation plane definition for each PMI object during automated authoring. If your team is moving toward model-based definition and away from 2D drawings, this is the sort of unglamorous automation that quietly saves hours, because placing annotation planes by hand across a large model is slow and inconsistent between one engineer and the next.
Both of these are real AI doing real work on geometry. Both are scoped to the model in front of you.
Siemens Split the Job Between NX and Teamcenter
The single most useful thing to understand about Siemens’ approach is that the company did not build one assistant. It built two, for two different problems.
NX Design Copilot handles the CAD session and the documentation behind it. Teamcenter Copilot handles the product data. Siemens describes the Teamcenter side as answering plain-language questions with sourced answers drawn from documents and data across Teamcenter, analysing documents to pull out requirements and performance data, navigating and analysing complex multi-domain BOMs, running impact analysis and triggering multi-step change workflows, grading requirement quality against standards, generating test cases against standards such as ISO 26262, building manufacturing BOMs and assigning resources, and producing quality audit summaries.
That is a substantially wider reach than the CAD-side copilot. It is also a different product, with a different licence, living on a different system.
The consequence for a design engineer is concrete. If you are in NX and you want to know why the bracket in front of you was revised three times last quarter, who approved it, and what else it is used on, the copilot in your CAD window is not the thing that answers. That information is PLM data. A different assistant reads it, and only if your organisation runs Teamcenter and has licensed the capability.
This pattern repeats across the industry, as we found in our comparison of what each CAD and PLM vendor actually ships, and the same split shows up in what PTC ships inside Creo.
Where Neither Assistant Reaches
Between the CAD session and the PLM vault there is a third space, and it is where a great deal of engineering time actually goes.
Consider what a working engineer needs on a normal Tuesday: the calculation sheet from the last programme that sized a similar shaft, a supplier datasheet sitting as a PDF on a network drive, the internal standard that says which bearing series is approved, a test report from two years ago explaining why the previous design failed, and CAD files from a legacy vault that was never migrated. None of that is in Siemens documentation. Some of it is in Teamcenter and some of it is not. Most teams also run more than one CAD system, so a single-vendor assistant covers a fraction of the estate.
This is the retrieval gap, and it is the problem Leo was built for. Leo is an AI assistant for mechanical engineers, trained on more than a million pages of standards, books and articles, that connects to an organisation’s full knowledge base: PDM, PLM, local and network directories, and ERP. Leo offers integrations with leading PDM and PLM platforms (SolidWorks PDM, Autodesk Vault, PTC Windchill, Siemens Teamcenter, Arena PLM, and others). The point is not to replace what Siemens ships inside NX. It is to sit above it, so that a question spanning the vault, the network drive and the standards library returns one grounded answer with a citation you can open and check.
That matters most for the knowledge that never got written down in the first place. We looked at capturing what your most experienced engineers know before they leave, and the retrieval problem is the same one wearing different clothes.
On security, the questions worth asking any vendor are the same. Leo is SOC-2 certified and GDPR compliant, no AI is trained on customer data, and customer IP stays protected.
How to Decide What to Rely On
A simple triage works better than a blanket policy. Before you ask a question, work out which system actually holds the answer.
If the answer is in Siemens documentation, ask the Design Copilot. Command discovery, workflow guidance and onboarding questions are what it is grounded to do, and it does them well.
If the job is repetitive geometric selection inside one part, use Select Similar Faces and Edges. Check the sensitivity setting before a bulk edit, because a match that is too loose will catch features you did not intend to change.
If the question is about a record, a revision, a change or a BOM, that belongs to Teamcenter Copilot, assuming your organisation runs Teamcenter and has the capability enabled.
If the answer spans several systems, predates the current vault, or sits in a document nobody has indexed, no built-in assistant will find it. That needs a retrieval layer over your own sources.
When you pilot any of these, pick one measurable task rather than running a general trial. Time how long it currently takes to find a released part that meets a given requirement, then measure the same task again afterwards.
The mistake worth avoiding is assuming that because an assistant appears inside your CAD window, it can see everything your CAD system can see. It sees what it was grounded on. For a fuller picture of working with an AI layer alongside NX, see what engineers should expect from AI for Siemens NX, and for the PLM side, making Teamcenter search work.
FAQ
Siemens, “Siemens brings AI copilot to NX”, Siemens newsroom, Designcenter NX summer 2025 release announcement.
Siemens, “AI-enabled design: what’s new in Designcenter NX”, Siemens Designcenter blog, December 2025 release.
Siemens, “Teamcenter AI, featuring Teamcenter Copilot”, Siemens Teamcenter product page.
Answers across NX, Teamcenter and more
See what an AI layer above your CAD and PLM systems actually answers
Leo reads your PDM, PLM, network drives and documents, then answers engineering questions in plain language, with a citation you can open and verify against the source.
Schedule a Demo →
#1 New AI Software Globally - G2 2026
Enterprise-grade security
Trusted by world-class engineering teams
