Your lab's design software can now propose a complete crown from your scan in minutes, with no technician moving the mouse. 3Shape says more than 5 million design cases have been downloaded from its AI service. The questions at the chair: will that crown fit, will it hit, and who checks it before it is milled?
This article covers how AI crown design works, the main tools, what 2024 to 2026 studies found, what FDA records say, and how to send scans that both software and technicians can use. For the wider picture, read Will AI Replace Dentists and Dental Technicians?
Key Takeaways
- Four 2026 systematic reviews, of 7 to 24 studies each, found AI design consistently faster; three judged AI single crowns comparable or better, mostly in vitro.
- Contacts are the weak spot: a 2026 review recommends technician refinement of contacts, occlusion, incisal guidance and esthetics, and rates AI inlays and 3-unit bridges below human designs.
- In 25 anterior cases, AI crowns deviated 290 µm in incisal path from a technician's design as generated, and 132 µm after a technician refined them.
- Input matters: one AI tool erred more from adjacent teeth alone than from a quadrant, and exocad asks for at least 4 teeth plus gingiva.
- FDA places dental CAD/CAM systems in Class II, exempt from 510(k) with limits. None of the 1,614 devices on FDA's AI-enabled list is crown design software.
Quick Answer
AI crown design software drafts a crown from the scan in minutes. In 2026 systematic reviews, AI single crowns were generally comparable with human designs for shape, margins and internal fit, and consistently faster to design. The weak spots are occlusal and proximal contacts, cusp anatomy, wear facets, anterior esthetics and bridges, so a technician still reviews and refines. For you, clean scans and detailed Rxs matter more.
How Does AI Crown Design Work?
The commercial tools studied so far use deep learning: neural networks trained on scans paired with finished crown designs. Generative models, such as generative adversarial networks (GANs) and transformers, build a new tooth for the gap from the adjacent teeth or the whole jaw, and some also use the opposing teeth (Broll et al., 2024). 3Shape describes the scan and the lab's design preferences as the prompt. In its account, the AI designs the anatomy, while conventional algorithms build the inside of the crown and enforce minimum material thickness.
What the software reads from your case
- The margin: detected by the software or defined by the technician.
- The adjacent teeth: the crown is sized and its contacts targeted against them.
- The opposing teeth and bite: the target for occlusal contacts.
- Design preferences: tooth library, contact strength, surface detail. exocad calls them "rough guidance parameters" the AI might not follow exactly.
More input helps, up to a point. In scans from 30 patients, Dentbird Crown given only the adjacent teeth produced larger deviations and more failed reconstructions than with a quadrant, while full-jaw data did not improve on the quadrant (Broll et al., 2025).
Which AI Design Tools Can Labs Use?
Below is what each vendor says on its official pages; we do not rank them.
3Shape Automate
Launched in April 2021 as a design service "exclusively powered by AI technology", first for single posterior crowns; anterior teeth followed in 2024. It now covers crowns, inlays, onlays, bridges, copings, nightguards and models, in as little as 90 seconds, with a reported design acceptance rate above 94%. Labs review and can edit each design, and pay only for those they approve.
exocad AI Design
Built into DentalCAD 3.3 Chemnitz and 3.2 Elefsina (build 9036 or later), paid for with exocad Credits. It covers up to three adjacent posterior crowns, copings or pontics and single anterior crowns, on dies, custom abutments or screw-retained. In exocad's example the cloud calculation took about 90 seconds. exocad says the AI tailors the design to the patient's anatomy, "including adjacent teeth and occlusion".
Dentbird Crown
A browser-based AI tool from Imagoworks of Seoul for crowns, implant crowns, bridges, inlays and onlays. The company says a crown takes one click, users can then modify it, and the condition of adjacent and opposing teeth in the scan "may influence the accuracy".
What Do Studies Show About AI-Designed Crowns?
Four systematic reviews appeared in 2026: 24 studies of commercial AI programs for restorations and occlusal devices (Revilla-León et al.), 22 studies of single tooth-supported restorations, all in vitro (Ladia et al.), 15 comparative studies (Limpuangthip et al.) and 7 in vitro studies (Lemos et al.). Two more 2026 studies, covered in Dental Crowns Then and Now, reported 25 to 50% shorter design time in a review of 17 studies and 8 to 15 minutes saved per design.
Morphology
Overall shape of single crowns was generally comparable with human designs, but head-to-head tests still favor people in places: on a typodont molar, conventional 3Shape design was more precise than two AI programs, whose crowns had altered cusp angles (Sim et al., 2026), and technician designs matched the original tooth best in another in vitro study (Broll et al., 2026).
Occlusal contacts
A recurring weakness, and not only for AI: in Broll's study no design group, technicians included, contacted all relevant antagonist teeth. In 23 patients, AI molar crowns had smaller occlusal contact areas than the natural teeth; aligning the design to wear facets on the opposing tooth improved static and dynamic occlusion (Chai et al., 2026). AI did not accurately reproduce functional cusp wear facets in another study (Wang et al., 2025).
Proximal contacts
AI contacts often needed refinement. In 3-unit bridges, two AI programs scored below certified dental technicians on occlusal and proximal contact areas and connector size (Sawangsri et al., 2026). In implant crowns, every automated method adapted proximal contacts less accurately than technicians (Ren et al., 2026). More in why crowns come back high or with open contacts.
Margins
On 100 prepared teeth, Automate matched technicians' acceptability for finish line detection, and the two AI programs completed 99% and 97% of designs (Sawangsri et al., 2025). In 62 patients with 3D-printed crowns, AI and conventional designs had equivalent buccal, mesial and distal margins but not lingual ones; the methods were equivalent where margins were well defined (Win et al., 2025).
Internal fit
A 2026 review found internal fit within clinically acceptable values, 55.4 to 83.1 µm (Lemos et al.), consistent with 3Shape's account that conventional algorithms, not the AI, shape the inside.
Design time
The reviews agree on speed: AI programs "consistently demonstrated shorter design times" (Revilla-León et al., 2026). In one study, working time with either AI program was a quarter of a CAD novice's (Wu et al., 2025).
Where technicians still step in
Average incisal path deviation of AI-generated anterior crowns from a technician's conventional design in 25 cases. After a technician refined the AI design, it fell to 132 µm (Çakmak et al., J Dent 2024).
Limpuangthip's review recommends that experienced operators refine proximal contacts, occlusion, incisal guidance and esthetics. Anterior crowns next to a diastema were also weak (Hlaing et al., 2026).
| Task | AI performance in studies | Human role |
|---|---|---|
| Finding the margin | Matched technicians on 100 teeth; lingual margins weaker in 62 patients | Verify every margin, especially subgingival |
| Overall anatomy | Generally comparable; technicians most precise in some tests | Adjust cusp angles and form |
| Occlusal contacts | Smaller contact areas than natural teeth; wear facets poorly reproduced | Set contacts to the real bite and wear |
| Proximal contacts | Often need refinement; below technicians on bridges, implant crowns | Set contact strength |
| Internal fit | 55.4 to 83.1 µm, clinically acceptable | Cement space, material thickness |
| Design time | Consistently shorter | Review time remains |
| Anterior esthetics | Incisal path off 290 µm as generated; weaker with diastemas | Guidance, proportions, patient wishes |
| Bridges and inlays | Less acceptable than human designs | Connectors, pontics, contacts |
What Does AI Design Change for Your Cases?
Your scan matters even more
3Shape says any result is based solely on the model's training and its input: the scan and design preferences. exocad asks for at least four teeth with surrounding gingiva, minimal holes or gaps, and at least one adjacent tooth, and warns that artifacts and braces can interfere. A technician who doubts a margin can call you; software cannot ask.
Turnaround
AI saves minutes on design, not on milling, sintering, finishing or quality checks. Ask your lab for its delivery time, not its design time.
Consistency
When five technicians designed the same two crowns, their occlusal surfaces differed by an average of 275.5 µm with conventional CAD and 25.7 µm with an AI program, Dentbird (Nagata et al., 2025). Consistent is not correct, though: software can repeat the same contact error on every case.
What AI cannot know
The software sees geometry. It does not know the shade or stump shade, the patient's esthetic wishes, whether the patient grinds, which occlusal scheme you want, or what the patient disliked about the old crown.
Does AI Crown Design Software Need FDA Clearance?
FDA regulates optical impression systems for CAD/CAM of dental restorations under 21 CFR 872.3661; such systems "may consist of a camera, scanner, or equivalent type of sensor and a computer with software." They are Class II with special controls and exempt from 510(k) premarket notification, subject to 21 CFR 872.9, which still requires a 510(k) for a different intended use or a device that "operates using a different fundamental scientific technology".
FDA's listing database shows Automate (3Shape) and Dentbird Crown (Imagoworks) under that product code, NOF, with no 510(k) number. Listing is not clearance: FDA rules state registration "does not in any way denote approval" of a firm's products. Custom abutment design software, by contrast, needs a 510(k): 3Shape's Abutment Designer Software was cleared in 2016 (K151455) and exocad's AbutmentCAD in 2021 (K193352). We found no FDA record or guidance addressing AI crown design specifically.
Checklist: Scans That AI and Technicians Can Both Use
- A readable margin all around: retract, dry, and capture structure beyond the finish line.
- At least the full quadrant, with teeth on both sides of the prep when present.
- The opposing arch, at least its quadrant, including wear facets.
- A bite scan in maximum intercuspation, checked on screen before sending.
- A clean file: no holes, floating fragments, cotton rolls or cord.
- A pre-op or provisional scan if you want the existing shape copied, with "copy" on the Rx.
- Flagged anatomy: diastemas, rotations, missing or extra teeth.
- An Rx with material, shade, stump shade, contact preference and functional notes, plus photos.
Export steps are in How to Send a Scan to a Dental Lab.
Digital Cases at Elegant Dental Laboratory
- Digital scans are reviewed before production.
- Every case is made in house by more than 35 technicians in our 7,000 sq. ft. Brooklyn facility, with no outsourcing.
- Standard turnaround is 5 business days on most restorative cases, with rush options available.
- Up to a 7-year warranty on final restorations, and up to 20 years on screw-retained implant abutments.
- Pickup by our own drivers in the New York area (call to confirm availability) and UPS shipping labels for practices anywhere.
Submit a digital case or call 877-335-5221.
Related Guides
- TRIOS vs. iTero in 2026: How the Two Leading Intraoral Scanners Compare
- Shade Matching for the Lab: How to Take Shade Photos and Send a Stump Shade
- The American Dental Lab, Then and Now: From Back Rooms to Digital Workflows
FAQ
Is an AI-designed crown as good as a technician-designed one?
For single crowns, 2026 reviews found AI designs generally comparable in shape, margins and internal fit, mostly in vitro. Contacts, anterior esthetics and bridges still benefit from technician refinement.
Will AI design get my crowns back faster?
It shortens design, not manufacturing. Ask your lab for its delivery time.
Can AI design bridges and anterior crowns?
Some tools offer both. In studies, AI 3-unit bridges were less acceptable than human designs, and anterior crowns needed incisal guidance refined.
Is AI crown design software FDA cleared?
Dental CAD/CAM systems are Class II and exempt from 510(k), with limits. Automate and Dentbird Crown are listed with FDA in that category without a 510(k) number.
Does AI choose the shade?
No. Design software produces the shape; shade comes from the material and finishing, based on your Rx and photos.
Sources (33)
- Revilla-León M, Barmak AB, Kois JC, Gómez-Polo M. Performance and Time Efficiency of Artificial Intelligence-Based Programs for the Automatic Design of Occlusal Devices and Restorations: A Systematic Review. J Esthet Restor Dent. 2026, online ahead of print. doi:10.1111/jerd.70271
- Ladia O, et al. Application of artificial intelligence in the design of single tooth-supported restorations: A systematic review. J Prosthet Dent. 2026;136(2):355-378. doi:10.1016/j.prosdent.2026.05.014
- Limpuangthip N, et al. Design quality and time efficiency of AI-based versus human-based design for tooth- and implant-supported fixed dental restorations: A systematic review. J Dent. 2026;170:106644. doi:10.1016/j.jdent.2026.106644
- Lemos LM, et al. Accuracy and time efficiency of deep learning-based method for single-crown design compared with a conventional CAD software program: A systematic review. J Prosthet Dent. 2026;135(5):e20-e28. doi:10.1016/j.prosdent.2025.10.046
- Holiel AA, et al. Accuracy and Functional Performance of Artificial Intelligence-Based Automated Crown Design Systems: A Systematic Review and Meta-Analysis. Biomed Eng Comput Biol. 2026;17. doi:10.1177/11795972261463709
- Gonçalves LFF, et al. In vitro evaluation of automated single-unit crown design using artificial intelligence. J Dent. 2026;173:106755. doi:10.1016/j.jdent.2026.106755
- Broll A, et al. Generative deep learning approaches for the design of dental restorations: A narrative review. J Dent. 2024;145:104988. doi:10.1016/j.jdent.2024.104988
- Broll A, et al. Morphological effects of input data quantity in AI-powered dental crown design. J Dent. 2025;159:105767. doi:10.1016/j.jdent.2025.105767
- Broll A, et al. Influence of digital crown design software on morphology, occlusal characteristics, fracture force and marginal fit. Dent Mater. 2026;42(1):8-15. doi:10.1016/j.dental.2025.09.003
- Sim JH, et al. Comparative evaluation of precision, morphology, and structural stability in AI-based dental crown design software programs. J Prosthet Dent. 2026, online ahead of print. doi:10.1016/j.prosdent.2026.07.012
- Chai X, et al. Assessing a method of optimizing digital dental crown design with antagonist tooth-wear facets: A prospective clinical study. J Prosthet Dent. 2026;136(2):438-445. doi:10.1016/j.prosdent.2026.02.012
- Wang Y, et al. Comparative analysis of full crown morphology designed by artificial intelligence and dental technicians. J Dent. 2025;163:106131. doi:10.1016/j.jdent.2025.106131
- Sawangsri K, et al. Acceptability, deviation, and efficiency of 2 fully automated CAD software programs in designing 3-unit fixed dental prostheses. J Prosthet Dent. 2026;135(4):e64-e74. doi:10.1016/j.prosdent.2025.11.034
- Ren S, et al. A deep learning system on monolithic implant-supported crown design: Evaluating AI-generated models against conventional software outputs. J Prosthodont. 2026;35(5):712-720. doi:10.1111/jopr.70067
- Sawangsri K, et al. Acceptability and deviation of finish line detection and restoration contour design in single-unit crown: Comparative evaluation between 2 AI-based CAD software programs and dental laboratory technicians. J Prosthet Dent. 2025;134(2):409-417. doi:10.1016/j.prosdent.2025.03.037
- Win TT, et al. Fit accuracy of complete crowns fabricated by generative artificial intelligence design: a comparative clinical study. J Adv Prosthodont. 2025;17(4):224-234. doi:10.4047/jap.2025.17.4.224
- Wu Z, et al. Comparison of the Efficacy of Artificial Intelligence-Powered Software in Crown Design: An In Vitro Study. Int Dent J. 2025;75(1):127-134. doi:10.1016/j.identj.2024.06.023
- Çakmak G, et al. Can deep learning-designed anterior tooth-borne crown fulfill morphologic, aesthetic, and functional criteria in clinical practice? J Dent. 2024;150:105368. doi:10.1016/j.jdent.2024.105368
- Hlaing NHMM, et al. Artificial Intelligence-Driven Automated Design of Anterior and Posterior Crowns Under Diverse Occlusal Scenarios. J Esthet Restor Dent. 2026;38(1):129-142. doi:10.1111/jerd.70029
- Nagata K, et al. Verification of the accuracy and design time of crowns designed with artificial intelligence. J Adv Prosthodont. 2025;17(1):1-10. doi:10.4047/jap.2025.17.1.1
- 3Shape. 3Shape Automate: AI-driven dental design service (indications, turnaround, prices, design count and acceptance rate). 3shape.com
- 3Shape. Meet 3Shape Automate: World's first AI powered design service for labs. Press release, April 7, 2021. 3shape.com
- 3Shape. 3Shape Unleashes Cutting-Edge AI Workflows for Unprecedented Efficiency in Dental Labs. Press release, February 22, 2024. 3shape.com
- Hollenbeck K (3Shape). Use generative AI for your dental crown designs. 3Shape blog, August 1, 2023. 3shape.com
- exocad. AI Design. exocad.com
- exocad. AI Crown Design (user documentation wiki). wiki.exocad.com
- Imagoworks. Dentbird Crown. dentbird.com
- Code of Federal Regulations. 21 CFR 872.3661, Optical Impression Systems for CAD/CAM. ecfr.gov
- Code of Federal Regulations. 21 CFR 872.9, Limitations of exemptions from section 510(k), and 21 CFR 807.39, Misbranding by reference to establishment registration. ecfr.gov
- US FDA. Product classification, product code NOF (510(k) exempt) and product code PNP, dental abutment design software (510(k)). accessdata.fda.gov
- US FDA. Establishment registration and device listing records for Automate (3Shape A/S) and Dentbird Crown (Imagoworks Inc.), via openFDA. api.fda.gov
- US FDA. 510(k) premarket notifications K151455 (3Shape Abutment Designer Software, 2016) and K193352 (exocad AbutmentCAD, 2021). accessdata.fda.gov
- US FDA. Artificial Intelligence-Enabled Medical Devices (list, content current as of September 4, 2026). fda.gov
Reviewed by Gary Fingerman, founder and CEO of Elegant Dental Laboratory, Brooklyn, New York (founded 2007). Last reviewed: September 28, 2026. Vendor descriptions and figures are the companies' own claims as published on their websites; study results come from the cited publications.
