Live Demo: AR Glasses from Raw CAD to a Batch of 80+ Simulations and a Design Risk Report in One Day

Smart and AR glasses pack a display, lenses, a speaker, circuit boards and a battery into a frame that is dropped, bent, hit and worn every day. Checking such a product against all its load cases usually means weeks of model preparation, runs and report writing for every design iteration.

In this live demo, recorded in one session, the FEA AI agent behind the AutoCAE.AI SaaS platform automatically takes an AR glasses CAD with real errors all the way to a design risk report -- with the engineer only reviewing and approving.

Watch the full live demo (3 minutes) on YouTube: https://youtu.be/64Ms2y3lwVo

The Starting Point: A CAD with Errors

The CAD is a complete pair of AR glasses as it comes from design: openings, uncleaned surfaces, duplicates, intersections and overlaps, plus logos and parts that are not needed for a structural model. Normally all of this is fixed by hand before any meshing can start.

Step 1: One Request, Automatic Repair and Meshing

The request names the CAD and asks for the pre-processing. From there the FEA AI agent behind the platform automatically cleans, repairs and simplifies the geometry, meshes the model and runs it through the full check list -- in batch mode in the background. A notification arrives when it is done.

The result after about 1 hour:

A model with 1.33 million elements that passed all checks.
DataCheck passed with zero errors.

Step 2: Review, Then Single Drop Validation

The engineer opens the model for a quick check and approves the next step. The single drop runs on the compute cluster, and the platform generates the single drop report and uploads it to the shared drive:

Outline, open items for the customer and the validation summary.
Product introduction: exploded view, possible design changes from the last iteration, BOM and mass bridge.
The reliability specs from the customer.
The animated drop from several views: no components flying out.
Balanced energies and low hourglass energy.
Peak strain and stress of the main parts.

All model validation checks passed. The model is now trusted for every other load case.

Step 3: The Plan of Attack Becomes a Batch

After the customer approves, one sentence turns the plan of attack into batch jobs: seven load cases with 59 sub cases. A check holds two load cases back before they waste compute time; after the fix, all seven run.

Step 4: One Design Risk Report

When all jobs are finished, every result goes into one risk summary report that lists every risky part, by load case:

Oriented drops -- every orientation, the failure probability of each part, and a strain envelope for each critical part that shows where it fails.
Ball impact at 22 locations -- the front lens cracks, but the ball never reaches the eye. Lens strain and eye distance are shown for every location.
Rub and buzz -- the speaker sweeps through all frequencies; one interface closes at 908 Hz, a buzz risk.
Stiffness -- two bending cases and torsion.
Natural frequencies and mode shapes.
Thermal-mechanical -- from -40 C to 80 C.

In total, more than 80 simulations: from CAD to the design risk report in one day.

Why This Matters

The engineer's time goes into the two decisions that need a human -- approving the model and approving the plan -- not into repairing geometry, preparing decks or copying pictures into slides. Every design iteration goes through the same automated workflow on the AutoCAE.AI platform: same model rules, same checks, same load cases, same report structure.

CAD in, design risk report out -- automated. Book a demo to see it on your own product.