BREAKING  Prototyping.io joins Y Combinator's Spring 2026 batch CAD to finished CNC part in as little as ONE DAY Mechanical-parts manufacturing is a $1 TRILLION market still quoting by hand Tolerances down to ±0.0002″, no minimum order quantity Serving robotics, defense, automotive & AI infra teams BREAKING  Prototyping.io joins Y Combinator's Spring 2026 batch CAD to finished CNC part in as little as ONE DAY Mechanical-parts manufacturing is a $1 TRILLION market still quoting by hand Tolerances down to ±0.0002″, no minimum order quantity Serving robotics, defense, automotive & AI infra teams
Company · Manufacturing × AI

Prototyping.io wants your CAD file to become a finished part by tomorrow

The San Francisco startup is automating the dull, expensive middle of manufacturing - the quoting, the planning, the machine programming - so hardware teams can iterate in days, not weeks.

Every hardware engineer knows the pause. You finish a part in your CAD software, feel briefly proud of it, and then send it out to be quoted. What comes back is not a part. It is a wait. Days, sometimes weeks, of a human somewhere reading your drawing, deciding whether the machine can actually make it, planning the cuts, writing the program, and eventually - if the tariffs and the shipping and the misread tolerance all cooperate - a box on your desk. Prototyping.io was built to shrink that pause down to a day.

The company, based in San Francisco and part of Y Combinator's Spring 2026 batch, describes itself in four plain words: autonomous manufacturing for mechanical parts. Underneath the plainness is a fairly stubborn bet - that most of the time between a design and a finished part is not spent cutting metal. It is spent on paperwork, judgment calls, and programming that a machine could, in theory, do itself.

"Turn CAD designs into CNC machined parts - driven by AI." Prototyping.io, company site

The problemA trillion-dollar industry that still quotes by hand

More than a trillion dollars is spent every year making mechanical parts. For a number that large, the process is remarkably manual. Before a single chip of aluminum comes off, someone has to run design-for-manufacturability checks, source the job, plan the sequence, program the machine, and set up the tooling. Each step is a place to lose a day. Domestic shops are expensive and fragmented, with machines that sit idle between jobs. Sending the work offshore trades that for shipping delays, tariffs, and a supply chain you cannot see into.

The result is a tax that lands hardest on the people who can least afford it - teams trying to iterate. In hardware, the team that gets to hold ten versions of a part usually beats the team that got to hold three. When each version costs weeks, iteration slows to a crawl.

$1T+Spent yearly on mechanical parts
~1 dayTarget CAD-to-part turnaround
±0.0002″Tightest stated tolerance

The productSoftware wearing a machine's clothes

Prototyping.io's insight is that the slow part of manufacturing is really a software problem in disguise. So the platform automates the path rather than the metal. Upload a CAD file and the system extracts its features, evaluates whether it can actually be made, decides the optimal process and machine, generates the machine program, and then delivers the finished part. The steps a shop foreman would walk through on a clipboard become, mostly, code.

How a file becomes a part
01
Upload
AI extracts features from raw CAD
02
DFM
Manufacturability checked early
03
Plan
Best process & machine chosen
04
Program
Machine code generated & run
05
Deliver
Finished parts shipped
CAD → DFM → PLANNING → SOURCING → PROGRAMMING → SETUPS → MANUFACTURING EXECUTION

The catalog behind that workflow is broad. CNC machining covers 3-, 4-, and 5-axis milling, turning, and EDM. Beyond it sit sheet metal, 3D printing, injection molding, extrusion, and die casting. There are more than forty materials and a range of finishes, and - notably for anyone who has ever needed exactly one bracket - no minimum order quantity.

Fast, reliable parts delivered through optimized, automated manufacturing workflows at lower costs. The company's own pitch

The foundersFrom AI software to the factory floor

Prototyping.io was started in 2026 by Revanth Bodepudi, who serves as CEO, and Prerit Oberai, the co-founder and CTO. Bodepudi studied at IIT Bhubaneswar and the University of Texas at Austin. Oberai came from years of building software, including AI systems at Microsoft, and studied at the University of Illinois at Urbana-Champaign. It is a two-person company at the time of writing, hiring its first operations and engineering roles out of the Bay Area. The pairing is telling: a team that treats a machine shop as something you can program.

The customersWho is actually ordering

The people reaching for this are hardware teams, and they range widely - from early-stage startups to multi-billion-dollar enterprises. The sectors read like a list of things that are hard to prototype: robotics, AI infrastructure, energy, automotive, and defense. According to the company, larger customers are already saving weeks on iteration cycles by routing prototype and early-production parts through the platform.

Iteration cycle, before vs. after
Illustrative - based on the company's "weeks to days" claim for hardware iteration.
Traditional
~weeks
Prototyping.io
~days

The businessUpload a file, pay for a part

The model is refreshingly legible. Customers upload designs and pay per part, made to order. The AI that handles quoting, DFM, planning, and programming is what lets the company promise lower cost and faster turnaround without a human touching every job. In its YC launch, the company reported roughly $400,000 in monthly revenue - a figure worth treating as an approximate, self-reported snapshot rather than an audited line. It raised a $130,000 seed as part of the batch, with YC partner Nicolas Dessaigne.

The fieldWhere it sits on the shelf

Prototyping.io is not the first company to offer instant online quotes for custom parts. Xometry, Protolabs, Fictiv, and the network formerly known as Hubs all sell some version of "upload CAD, get parts," alongside thousands of local and offshore shops that still do it by phone and PDF. What Prototyping.io is arguing is that the layer everyone treats as overhead - the DFM, the planning, the CNC programming - is exactly the layer AI should own. Win that boring middle, the bet goes, and the speed and cost advantages follow the whole way down.

The best place to insert AI is not the flashy front end. It is the invisible middle that nobody wants to do. The wager underneath the pitch

Whether that thesis holds is the open question. Autonomous execution - the industrial robotics and highly automated production the company describes as its longer arc - is far harder than automated quoting. But there is something clarifying about the goal. If getting a machined part made could feel as ordinary as ordering cloud compute, a lot of hardware would get built faster. Prototyping.io is trying to make the distance between a design on a screen and a part in your hand as short as the technology will allow.

#manufacturing#ai#cnc-machining#hardware#dfm#cad-to-part#yc-p26#physical-ai#rapid-prototyping