GERMANY VIDEO REVIEW · WHERE TO AUTOMATE
Which Parts Should You Automate First in CAM? A German Reseller's 38-Minute Webinar Has the Answer
01What this webinar is about
CINTEG is a German CAD/CAM systems integrator (a hyperMILL partner). This webinar, recorded in 2022, runs a full 38 minutes on a single theme: "Automating CAM programming with hyperMILL"—and its description opens by noting that many companies have already used it to automate NC programming and successfully accelerate their workflows[1].
For Taiwan readers, the most valuable thing in this video isn't the tool operation (that's for hyperMILL users), but the methodology it opens with—which parts to start automating from.
02The automation-entry-point list: families, features, workflows
The official description spells out the "good entry points (gute Ansatzpunkte)" clearly[1]:
| Entry point | Original German | Why it suits automation |
|---|---|---|
| Part families and variants | Teilefamilien und Varianten | Parts in a family differ only by dimensional parameters; one template covers the whole family |
| Recurring shape elements | wiederkehrende Formelemente | The same geometry gets the same strategy; the rule is written once |
| Features: holes, pockets | Features wie Bohrungen oder Taschen | Features can be recognized → strategies can be assigned automatically |
| Repetitive workflows | sich wiederholende Abläufe | Work with a fixed process is the best fit for scripting |
Note that the flip side of this list is just as useful: what isn't on the list (one-off odd shapes, brand-new processes, experimental projects) is not the starting point for automation—that's a senior engineer's home turf. Automate the high-repetition 80% first, concentrate people on the low-repetition 20%—the shared structure of every successful rollout.
03The theoretical basis: group technology and feature-based planning
Behind this list are two classic theories:
- Group Technology: group parts of similar geometry and process into families that share a process plan—automation textbooks list it as one of the foundational methods of CAPP, and its benefit is exactly "plan once, reuse across the family"[3].
- Feature-based planning: features like holes, slots and pockets are the natural unit of correspondence from "geometry → operation"; feature-recognition research (e.g. FeatureNet reaching about 96.7% benchmark recognition accuracy with deep learning) lets this correspondence be built automatically[4], and the CAPP review literature likewise lists feature recognition as the first step of automated planning[2].
In other words, what this webinar teaches isn't a trick unique to hyperMILL, but a universal principle that applies in any CAM environment—the tool changes, the methodology doesn't.
04The audit checklist for Taiwan shops and the AI bonus
Following the webinar's method, a Taiwan shop can run a self-audit right away: lay out the last six months of orders and ask four questions—how many jobs belong to the same part family? Which features (hole patterns, pockets, threads) recur? Which workflows repeat on every job (drawing reading, modeling, program sheets, first-article reports)? And what share is left that is genuinely "different every time"? Most job shops will find that the automatable share is far higher than intuition suggests.
And AI drawing-reading adds a layer of capability this method never had before: the task of "finding similar parts" is itself automated. Traditional group technology required manually building classification codes; in the course of AI reading the drawing and building the model, features and geometry are already structured—the system can classify the part spectrum on its own and discover that "this new drawing is the same family as that part from three months ago" (for a German enterprise-grade implementation of the same idea, see the similar-workpiece search in the OPEN MIND piece). BestAI CAM's prep chain works exactly this way: AI reads the 2D drawing, recognizes features, builds the 3D model, and generates code per the shop's tool library and machine limits, with simulation and dimensional cross-check—high-repetition parts take the fully automated path, odd shapes take an AI-accelerated semi-automatic path, and the part-spectrum data accumulates automatically with every job (see The Complete Guide to CNC Auto-Programming).
05FAQ
Which parts should you start CAM automation with?
Per the webinar's checklist: part families and variants (same family, only parameters differ), recurring shape elements, recognizable features such as holes and pockets, and the workflows that repeat on every job. One-off odd shapes and brand-new processes are not the starting point—automate the high-repetition bulk first, and concentrate senior talent on the few that genuinely need judgment.
Our parts all look different from each other—does this apply to us?
We suggest auditing six months of orders before concluding: group them by geometric similarity into families, and tally recurring features and repetitive workflows. Most job shops have far more actual repetition than intuition suggests—"every part is different" is often only the outline, while hole patterns, pockets, fixturing and workflow repeat heavily. AI drawing-reading tools can even automate this audit itself.
Group technology is an old method—does it still apply in the AI era?
It not only applies, AI has filled in its most expensive part. Traditional group technology got stuck on manually building and maintaining classification codes; AI drawing-reading structures features and geometry automatically, so the system discovers and classifies similar parts on its own. The methodology is unchanged (similar parts share a plan), but the execution cost drops sharply—which is exactly why the old theory pays off again in the AI era.
We'll let you know when new articles and video walkthroughs go live—no inbox flooding, one-click unsubscribe. (newsletter in Chinese)
AUTOMATE THE REPEATING 80%
Audit your part spectrum first, then let AI eat the high-repetition stretch
AI reads drawings and auto-classifies the part spectrum, family parts run fully automatic code generation, odd shapes get semi-automatic acceleration—audit once with a few representative drawings.
Contact an onboarding consultant Training courses06References
- CINTEG AG (YouTube). CAM Programmierung automatisiert mit hyperMILL (published 2022-03-14, 38-minute webinar). youtube.com/watch?v=zBiYrCqMQJ8
- Xu, X., Wang, L., & Newman, S. T. (2011). Computer-aided process planning — A critical review of recent developments and future trends. International Journal of Computer Integrated Manufacturing, 24(1), 1–31.
- Groover, M. P. (2019). Automation, Production Systems, and Computer-Integrated Manufacturing (5th ed.). Pearson.
- Zhang, Z., Jaiswal, P., & Rai, R. (2018). FeatureNet: Machining feature recognition based on 3D Convolutional Neural Network. Computer-Aided Design, 101, 12–22.