Pricing on a custom-manufactured part is a small machine learning problem wrapped in a large heuristics problem. Geometry varies, materials vary, supplier capacity varies. Anyone claiming a single magic formula is hiding things. This article shows how the Uretyco quoting engine actually works, step by step.
Step 1: Geometry analysis
The moment a STEP, STL, or 3MF lands on the platform, it is meshed and analyzed for: bounding box, volume, surface area, hole count and depths, internal corner minima, deepest pocket aspect ratio, thread features, undercut detection, and thin-wall regions. These metrics drive every downstream decision.
- Bounding box determines stock size and machine fit.
- Volume drives material cost.
- Surface area influences finishing time.
- Pocket depth and corner radii drive cutter selection.
- Undercuts and threads trigger additional setups.
Step 2: Process suitability
We score every part against each manufacturing process — CNC milling, CNC turning, sheet metal, FDM, SLS, MJF, injection molding, casting. The score considers feasibility (is this geometry possible?), economy (is it cost-competitive?), and lead time. Multiple processes are surfaced when the choice is genuinely a tradeoff.
Step 3: Material selection
The default material follows the most common selection for that process and geometry, but the AI co-pilot proposes alternatives based on the use case description. If you tell us the part is for an outdoor enclosure, the engine will recommend 6061 with anodize over generic mild steel.
The platform's recommendation is just that — a recommendation. Every quote shows the rationale. You can change material, finish, tolerance class, and quantity, and the price updates immediately.
Step 4: Machine time estimation
Machine time is estimated using a hybrid model: a physics-based estimate (volume removed × removal rate, plus tool change overhead) calibrated against the historical actual times reported by suppliers. The model gets better with every order.
| Cost component | Typical share of CNC quote |
|---|---|
| Material stock | 10 - 30% |
| Machine time | 35 - 55% |
| Setups & fixtures | 10 - 20% |
| Inspection & deburring | 5 - 10% |
| Finishing (anodize, paint) | 0 - 25% |
| Logistics & packaging | 5 - 10% |
Step 5: Supplier matching
Once the cost model produces a price, we check which suppliers in the network can deliver this part with the required quality and lead time. Suppliers are ranked by: process and material capability, current load, historical on-time rate, defect rate, and proximity to the destination.
Step 6: Rationale and trace
Every line on a Uretyco quote can be expanded to show why it was added. Setup count, why the chosen material, what drove the lead time. We do not believe in opaque pricing — and we have found that customers who understand the breakdown design better, faster.
What the AI does and does not do
Our AI co-pilot suggests material and process choices, flags DFM concerns, and estimates risk. It does not autonomously edit your design, change quantities you specify, or sign contracts. The model is trained on anonymized quote and outcome data — no customer CAD ever leaves the platform for training purposes without explicit, granular consent.
Where the engine is honest about limits
Very large parts (>1 m), exotic alloys (Inconel, Hastelloy), and sub-micron tolerances trigger a manual review path. The platform tells you upfront when this happens; you get a confirmed quote within one business day instead of an instant number. Honesty about edges is part of the product.
Related articles
Can't find what you need? Contact our team for expert assistance with your manufacturing project.