Is your part mold-ready?
Know in seconds.
An AI trained on 28,905 industry-standard fill simulations (the 2.5D thin-wall method behind Moldflow's midplane and dual-domain analyses) of 3,979 real-shaped parts. Upload a thin-walled part: once it is meshed, every gate and setting you try gets a simulation-grade fill prediction in under a second, checked by physics right in your browser.
Opens the tool. Your file never leaves your browser.
Demo video coming soon
Time to one fill prediction
100–1,000×faster than Autodesk Fusion's cloud simulation, file to answer, including the queue
This AI, once meshed< 1 s
Its physics check½–4 min
Autodesk Fusion cloud15–60 min
Log scale. Fusion: 56 runs in this project (Oct 2026), typically 15–60 min per run including the queue; a local desktop Moldflow run can take only a few minutes for small parts. AI and physics are timed live in your browser. Uploaded STEP files are meshed first, which takes from about half a minute to a few minutes in the browser.
Cost per year
$390MoldReady Pro (planned)$3,000–47,000conventional mold-flow software
List prices found Oct 2026, may vary; sources and plans below. Those tools also simulate packing, cooling and warpage.
In shortFour neural networks learned from 28,905 physics simulations of 3,979 parts. They predict the fill in under a second, and the same physics re-checks it in your browser.
6%Our physics is typically within 6% of Autodesk Moldflow inside its validated range (median of 18 test runs; most thin-walled consumer parts)
3%The AI vs our physics: median fill-time error on 1,708 test runs it never saw
Limits
Validated for thin walls up to 2.5 mm. Outside that, answers are marked amber and shown for guidance.
Full limits
- Validated range: walls up to 2.5 mm, no cold, fast fills (under 220 °C and under 0.75 s) on walls over 1.5 mm, not near freeze-off. Inside it the physics is typically within 6% of Moldflow (2 in 3 runs within 10%).
- Known gaps inside the range: small flat plates and plates with slots or many holes can read 20–40% low.
- Outside the range: typically ±20%, possibly because Moldflow solves thick sections in 3D, a known limit of 2.5D. Across all 30 Moldflow test runs the physics is within 14.4%.
- One known AI outlier: one tall sleeve in our Moldflow test set reads about 45% high on pressure; the physics check shows the right value.
- Speed: the AI runs on your graphics card through WebGPU, under 0.1 s per prediction. On browsers without WebGPU it runs on the CPU: about 1.5 s for a 3,900-point remote cover up to 5 s for an 11,600-point food container. The physics check takes half a minute to a few minutes, depending on part size.
- Scope: the fill stage only, no packing, cooling or warpage. The win is speed at comparable accuracy, not more accuracy than Moldflow.
The numbers in detail
The AI is scored against the 2.5D physics solver it learned from, on parts and settings it never saw in training; the solver itself is scored against Autodesk Moldflow (Fusion 360). Not yet against molded parts.
3.0%median fill-time error: about 2.3% on generated boxes and trays, 3–4% on real CAD parts
7%median error in the peak injection pressure
9 in 10short-shot areas caught by the AI; the automatic physics check covers the rest
for one AI prediction in your browser, measured just now
for the same fill with physics in your browser, measured just now
How it works
The slow part is done once. After that every answer is a quick pass through the network.
Done once, before you opened this page
- Simulate
- Train
- ShipThe four trained networks, about 5.6 MB of numbers, load with this page. So does the solver itself.
Happens live, every time you change something
- Describe the partYour part is meshed into triangles about 2.5 mm across on both faces of every wall, and the wall thickness is measured between them. Each mesh point knows its thickness, its neighbours and its distance from the gate.
- AI predictsThe network passes information between neighbouring mesh points 12 times, then outputs fill time, fill or no-fill, pressure, frozen layer and temperature for every point. Clamp force and weld lines are worked out from that. It runs on your graphics card (WebGPU): about 0.02 to 0.1 s per prediction on a laptop. On browsers without WebGPU it runs on the CPU instead, about 1 to 5 s depending on part size.
- Physics checksThe 2.5D thin-wall method used by commercial tools (generalized Hele-Shaw): real viscosity data for the plastic, temperature through the wall thickness, a frozen layer that grows from that temperature, and a set fill time with a pressure limit. It is the same Python solver that made the training data, running in your browser.
- CompareA badge above the 3D view says how closely the AI matched the physics; the AI vs physics tab shows the two side by side.
Planned pricing
Proof of concept; not on sale yet. The fast fill check you run first, before paying for full simulation or cutting steel.
Free
$0Students and makers- 5 parts a month
- AI + physics check
- Watermarked summary
Pro recommended
$39/monthor $390/year · designers and small companies- Unlimited checks
- Gate optimizer
- Toolmaker summary (PDF)
Team
$1,500/year5 seats · product teams and toolmakers- Shared projects
- More materials
- Priority support
Single report
$19/partOccasional users- One part, full check
- Toolmaker summary (PDF)
8× cheaper than entry-level, ~25× cheaper than Moldflow Adviser, ~100× cheaper than Moldflow Insight.
Those tools also simulate packing, cooling and warpage. MoldReady is the fast fill check you run first, before paying for full simulation or cutting steel.
See the price comparison
| Tool | Price per year | Notes |
|---|---|---|
| Autodesk Moldflow Insight | $35,000–47,000 | CDW list: Premium renewal to Ultimate, new |
| Fusion with Moldflow Adviser | ≈ €9,350 / NZ$13,345 | German list price (May 2026) / Autodesk NZ |
| SolidWorks Plastics | $2,998 / $8,248 / $12,373 | TriMech: Standard / Professional / Premium; needs a SolidWorks licence |
| Fusion + Simulation Extension | $680 + ≈ $1,465 | Autodesk list price |
| Outsourced analysis | ≈ $180 per part | Service-bureau quote |
| MoldReady Pro (planned) | $390 | Fill check only |
List prices found Oct 2026; may vary. CDW: Moldflow Insight Ultimate, new CDW: Moldflow Insight Premium, renewal ki-syndikat.de: Autodesk Moldflow Autodesk NZ: Moldflow TriMech: SolidWorks Plastics Autodesk: Fusion subscription FAQ Autodesk Fusion blog: Simulation Extension IndiaMART: mold-flow analysis service
From proof of concept to product
The demo proves the idea on simple thin-walled parts with one plastic. These are the steps between it and a tool a mold shop could rely on.
- Any partThick and solid sections too, with full 3D flow, not only thin walls.
- More materialsMeasured data for grades of ABS, PP, PC and others, beyond the one polystyrene used now.
- ValidationMore parts against commercial simulation, closing the flat-plate and cold-fill gaps, and measured filling of real molded parts.
- The full cyclePacking, cooling and warpage, not only the fill.
What this prototype simplifies
- Thin-walled parts only: walls about 1 to 4 mm, up to 150 mm across. Thick, blocky or closed parts are rejected. The method is 2.5D: the melt flows along the walls with a temperature profile through their thickness, so 3D flow in thick sections and right at sharp corners is not modelled.
- Filling only. No packing, cooling or warpage stage. Clamp force uses the pressure when the cavity is 99% full on the projected area, and real machines need more because packing pressure is higher.
- One plastic: polystyrene Styron 678E, with the Cross-WLF viscosity and thermal data published by Vietri et al. (2011), which match the Moldflow database entry for this grade, as distributed with openInjMoldSim. The viscosity's shear-thinning and temperature terms (n, D1, A1) are fitted to Autodesk Moldflow (Fusion 360) on calibration strips only, the pressure dependence is left out (D3 = 0, the database value), the mold-to-melt heat transfer coefficient is 1,500 W/m²K, and drag from the side walls near free edges is included.
- How the physics was checked: against exact solutions for strip and disk flow, mass conservation, mesh refinement, heat conduction and an energy balance; and against measured solidification times in a Styron 678E strip (Speranza, Vietri and Pantani 2013), within about 20% at 200 °C, while at 220 to 240 °C it underestimates by about 40% because packing flow is not modelled. Against Autodesk Moldflow (Fusion 360, same plastic and settings, 33 runs; parameters were fitted only on separate calibration strips and 3 calibration runs): injection pressure at switchover within about 6% on average on boxes, trays, a lid and a housing (12 runs), and inside the validated range a median of about 7%. Across all 30 non-calibration runs the error is 13.5% on average (median 12%). It reads low on thick walls (a 3.5 mm housing about 25–29% low), small or flat parts (a 36 mm plate −32%, a slotted plate −12 to −30%) and cold, fast fills (−7 to −30%). No measured filling pressure was available.
- The AI's accuracy figures above are measured against this solver on held-out parts, not against a commercial tool or real parts.
- The gate search ranks every candidate position with the AI, using a simple score we picked by hand (short shot first, then the injection pressure the fill needs, then weld lines). It is not a validated measure of part quality. Physics is optional: any pick can be confirmed with a full solve (half a minute to a few minutes), or every gate checked at once, which takes much longer.
- The Build tab makes simple open boxes, trays and plates (rectangular, rounded or L-shaped, with round holes and vent slots); ribs, bosses, rims and thin panels are not in it yet. Uploaded STEP files keep their real shape.
Terms used on this page
- Gate ★
- Where molten plastic enters the mold. Its position changes everything about the fill.
- Fill time
- When the melt reaches each spot, from 0 s at the gate.
- Short shot
- Plastic freezes before reaching some areas, so the part comes out incomplete.
- Weld line
- Where two melt fronts meet. Often a visible line and a weak spot.
- Air trap
- A pocket the melt surrounds with no edge for the air to escape. Can cause burns or voids.
- Frozen skin
- Plastic freezes inward from the cold mold walls. The more of a wall that has frozen, the harder it is for melt to flow through.
- Clamp force
- How hard the machine must hold the mold shut against the melt pressure. Sets the machine size.
- Mesh point
- A corner of the triangle mesh, about 2.5 mm apart on both faces of each wall. The AI and the physics both give one value per point.
MoldReady. A proof of concept by Ryan Sehgal. © 2026 Ryan Sehgal. All rights reserved. No copying, reverse engineering or redistribution without written permission.Open the tool