Cal Hacks 13.0, team of 3
ForgeAI
Engineers can design a part in an afternoon, then wait days to find out what it costs to make.
Problem
One of our teammates is a mechanical engineer who kept hitting the same wall: design software knows everything about a part's geometry and nothing about its cost. Stock had to be sized by hand, machining time estimated, quotes chased, and the total checked against a budget that was already partly spent.
Users
Mechanical engineers and small manufacturing teams who need to know what a part will cost, and whether the budget can cover it, before they commit.
My role
One of a team of 3 at Cal Hacks 13.0, with Vincent Ly and Tao Tokikawa.
Approach
- Started from a teammate's real, repeated pain instead of a technology looking for a use.
- Cut a huge first feature list (supplier search, approval workflows, banking integrations) down to one question: what does this part cost, and can we afford it? Integrations that didn't serve that got cut, even ones that worked.
- Kept the hard research problems, like predicting machining time, as explicit, editable inputs with labeled assumptions rather than overclaiming.
- Added mesh checks and a fallback 3D viewer so a bad file or a machine without WebGL warns the user instead of silently showing a wrong number or a blank box.
Solution
Upload an STL and ForgeAI measures the part, prices it for CNC milling or FDM printing, and shows how unit cost falls with quantity. It checks the order against your balance, upcoming commitments and a protected reserve, tells you the largest quantity you can afford, then generates itemized invoices read aloud with Cartesia's voice AI.
Results
- A complete flow from STL upload to price, budget decision and spoken invoice
- Invoice line items trace back to the part's real geometry, not hard-coded numbers
- Cost model works in whole cents and is covered by unit tests
- Every estimate shows its assumptions, and the app labels what is measured versus simulated
What I'd do next
Connect the voice agent to a phone line so customers can confirm an invoice in a live call, and replace assumed rates with live supplier quotes and lead times.
