Seed proposal · $50 million

Embraxis
Robotics

Modular Robots for Disassembling Salvage Cars

The pack goes to storage. The parts go to the garage.

Salvage cars come in, modular robots extract the battery pack, and graded packs are sold for storage

What We DoWe build robots to disassemble salvage cars.

We grade and sell the battery pack and the parts to energy storage solutions company and Ebay resellers.

Product and Process

A small disassembly cell with ten actuators. Each one carries a single tool: drill, reciprocating saw, hex nut, grip, impact, pry, shear, Torx, clip, and probe.

Process

We buy the car for $800-2000

We disassemble it

grade parts and battery pack

Sell the parts

to storage companies and ebay resellers

The team

Name

Manipulation and learning

Name

Mechatronics and controls

Name

Battery safety and the line

The Market

Market Validation


Recyclers already do the bay

$32 billion in sales by selling parts from 4 million salvaged vehicles

LKQ

LKQ

Pull-A-Part

Pull-A-Part

Pick-n-Pull

Pick-n-Pull

Garages already buy the part

$12 billion in parts bought by garages

Caliber

Caliber

Gerber Collision

Gerber

The battery pack is already used

Storage companies buy used batteries from vehicles

B2U Storage Solutions

B2U

Moment Energy

Moment

Smartville

Smartville

Recycler counts are from Automotive Recyclers Association. Used-parts wholesale is analyzed in IBISWorld.

Competition

 

 

The process codified on a robot

Process starts over with a new car

Manufacturing new parts

Sell recycled parts in the market

Path Robotics

Path

GrayMatter Robotics

GrayMatter

Formic

Formic

Machina Labs

Machina

Vention

Vention

Intrinsic

Intrinsic

Embraxis

Embraxis

Car enthusiasts and owner-operator garages

LKQ

LKQ

Pull-A-Part

Pull-A-Part

Pick-n-Pull

Pick-n-Pull

Technicians, at $150–250 an hour, sit with the hired cells. Those six are a project, and the next car starts over. Embraxis buys the car, sells the pack and the parts, and the slope stays here.

Baby Steps to Bigger Market Opportunity

01 Car disassembly

The robot takes the salvage car apart. Older cars yield the parts. Electric cars yield the graded pack. The lab learns the fasteners here.

Parts and packs

02 Car repair

The same robot closes the car it opened. Buy the salvage car, fix it, and sell the car.

The car

03 Industrial maintenance

The proof leaves the bay for plants worldwide: ships, pulp and paper, defense manufacturing, data centers, and the rest of the work.

Worldwide

The lab will be situated in Marysville, Ohio., where a lot of manufacturing and industrial companies reside. Salvage cars are cheap in Ohio.

TAM, SAM, SOM

United States Worldwide
The plants. Maintenance and repair where the injury and the downtime cost the most $60 billion. Machinery and heavy-equipment repair, 2025 $142 billion in maintenance services, 2024. $290 billion with parts and supplies
TAM. Salvage cars, the first of those bays $32 billion in recycling sales. 12–15 million cars a year $116 billion in recycling sales, 2024. About 40 million cars a year
SAM. Parts a garage will buy $12 billion used-parts wholesale $17 billion recycled parts, 2024
SOM. One line in Marysville $8–40 million. 1,000–5,000 cars, then the same line in a plant $8–40 million. The same line, until a second plant

The plant row is the work these missions move into: ships, pulp and paper, defense manufacturing, and data centers. US repair is IBISWorld, 2025. Worldwide maintenance services are Grand View Research, 2024. Industrial MRO, including parts and supplies, is $290 billion. Ship repair, about $36 billion worldwide, is one of those high-injury trades. Cars are the bay that teaches the motion: ARA, Stratistics MRC, IBISWorld, and HTF. SOM is ARA’s $32 billion over about 4 million vehicles, about $8,000 a car, so 1,000–5,000 cars is $8–40 million.

Go to market

Storage

  • The first check. Marysville sells the graded pack direct
  • Data-center backup and peak shaving are the use
  • This round leaves with that offtake in motion

Garages

  • The parts counter for the older car, and later the repaired car
  • Enthusiast shops already buy used parts and sell the build
  • They are accounts, and the counter the creator sends a buyer to

Creators

  • A large YouTube or Instagram following brings the buyer to that counter
  • The film is the teardown and the fix, including a hard job such as an Audi piston skirt
  • The creator is the channel. The garage takes the order

The storage company is the customer of this round. The garage sells the parts. The creator brings that buyer.

Raising the seed round for $50 million for 20% stake

Seeking the Seed Round: $50 million

Use Amount What it stands up
Ohio factory, two cells $18M A teardown and repack line for EV packs, plus a rack and enclosure cell
Robots, tooling, sensing $12M Cells for disassembly, grading, deburring and finishing, welding, and CNC tending
Core team $12M Manipulation and RL, mechatronics and controls, battery safety, operators
Safety, feedstock, certification path $5M High-voltage handling, fire systems, first salvage inventory
Data engine and compute $3M Wearables on the line, on-site training, the shared model

Proposed allocation for this raise, $18 + $12 + $12 + $5 + $3 = $50 million. Most of the round is the factory and the robots. Compute stays a thin slice on purpose: the bottleneck is contact and changeover, not a larger pretrained model. Planning figures, to be rebuilt against a site and a bill of materials.

Derisking Strategy

01 Stand up

Ohio factory. Pack line plus rack cell. CDME as the testbed beside it.

First

02 Robotize

Disassembly, grading, finishing, welding, and CNC tending.

Then

03 Measure

Hours per new part, interventions per shift, and the slope across the first 50.

Proof

04 Sign

Offtake for storage and racks. Three to five Ohio design partners.

Out

Seed does not try to be the platform. It tries to make the next part cheaper than the last one, inside a factory whose products a buyer already wants.

Funding Milestones towards General Purpose Platform

Seed

  • Prove setup cost falls on hardware we own
  • Retire the embodiment gap on our own cars
  • First offtake is a storage company for the graded pack

Series A

  • Revenue from storage systems and racks
  • The platform in 10–20 external plants, priced per hour or per part
  • Onboarding time falls with each new site. Certify the battery product

Series B

  • Assembly, kitting, and inspection on the same model
  • Midwest, then defense and grid manufacturing
  • License the model. Finance fleets against assets, not against headcount


What has to be true first Three risks this round is sized to face, not to narrate away

Embodiment

  • Wearable hands are not robot hands
  • Transfer is the first research question
  • No external deployment until it moves

Battery safety

  • High voltage, fire, and certification
  • A safety group and a real lab
  • Feedstock will vary. Grading is a robot task

Focus

  • A factory and a lab in one building
  • Two cells and five tasks until the slope is real
  • Check any partner exclusivity against hyperscale lines before signing

The ask · $50 million seed

Robots that learn a new part in hours

Marysville builds IP buys the salvage car, sells the graded pack to a storage company and the parts to a garage, and keeps the learning curve. The same robot then does the plant.

  • Fund the factory, the robot cells, and a team that can run a line and a lab together.
  • Prove that wearable data can become a robot policy, and that a task can climb from about 85% toward 99%+ on our own cars.
  • Leave the round with a storage offtake in motion and a learning curve across the first 50 parts.

abhishek@embraxis.ai · +1 (217) 819 6382 · embraxis.ai

Day-to-day Workflow

One car

Older salvage car Electric salvage car
We buy The car at auction The car at auction
We sell Parts to the garage and the reseller The graded pack to storage, and the remaining parts
Industry sales About $8,000 a car The pack is the larger ticket. Parts sit on top
Our spread Auction price against the parts Auction price against the pack and the parts
This round records Auction price, parts invoice, hours on the car Auction price, pack invoice, parts invoice, hours on the car

$8,000 is ARA’s sales per dismantled car, from $32 billion across about 4 million vehicles. It is the industry’s revenue. Our margin is the spread, once Marysville records the auction price and the sale. One thousand to five thousand cars is the $8–40 million SOM.

What leaves the building

Line What leaves the building What the robots learn
Lead. Graded packs Salvaged packs, taken out, graded, and sold for stationary storage Unbolt, isolate, drop the pack, grade, and test
Second. Structures Racks, enclosures, cable trays, containment Cut, weld, deburr, and assemble
Same cells, later The ship, the mill, the defense shop, the data center The missions already learned on the car

The pack is the first product, and it is why a storage company is the check on an electric car. The rack cell is the second product, and it is why this raise stands up two cells. The plant work uses those same motions.

Prove that setup cost falls

Baseline

Fingers, fixtures, programming, and calibration for one new part

Weeks

per part

Target

Onboard the part on our line, then do the next one faster

Hours

and falling

Research bar

Take a task from about 85% to 99%+ in hours, with a recovery when it misses. A plant ultimately needs about 99.9%. Battery work is held to a high-voltage and fire-safety standard the whole time.

The learning curve

50

first parts and tasks, tracked on our own factory floor

What gets recorded

Hours to onboard a new part.
Interventions per shift.
The slope across those fifty.

The slope is the company.

Competitive Advantages

Our own cars

  • The demonstrations, the misses, and the retries stay on hardware we own
  • A plant is not the experiment

The bay already has a buyer

  • The pack and the parts are sold before a plant writes a check
  • A hired cell bills a deployment. This line sells what it took off the car

The next part costs less

  • One policy, shared across the cells
  • Hours to the next fastener is the scoreboard, and it has to fall across the first 50

A buyer can hire a crew for one task today. The advantage is the fiftieth part, on a car we bought.

WhyWhy Robot for Cars?

Robot for cars: Warranty Claims Eating the Bottomline

Labor

  • Skilled car technician’s time is spent on complex disassembly and assembly (non dexterous tasks), rather than fixing the issue
  • Typical cost to customer is $150-250 per hour

High mix of parts

  • Disassembly and assembly requires parts of different sizes
  • Every new part needs new tools and corresponding logic

Garage Owners and Car Ethusiasts

  • Want easily repairable robots so the work never stops
  • A preference to lease or pay per use
  • Customers’ want quick turnaround on their repair, and robots can disassemble and assemble overnight


2024 claims Cash each company paid to settle warranty work, numbers taken from its annual report

Mercedes-Benz

  • €3.62 billion in claims paid
  • 3.0% of vehicle sales
  • The bill rose 7% from 2023

BMW

  • €3.22 billion in claims paid
  • 2.5% of vehicle sales
  • The bill rose 20% from 2023

Ferrari

  • €61 million in claims paid
  • €6,715 set aside per car sold
  • The bill rose 21% from 2023

Disassembly and Assembly does not require dexterity

The technician’s hour

  • The robot unbolts, unclips, and drops the pack
  • The technician keeps the diagnosis and the fix
  • The car is ready for technician’s work when the shop opens in the morning

The next fastener

  • One set of hands, and a policy that learns the next clip
  • Wearables record the job on the car in front of us
  • The scoreboard is hours to the next part, on hardware we own

The check

  • A storage company buys the graded pack
  • A garage buys the parts
  • A plant buys the hour after Marysville has the slope

The warranty bill shows the cost of the labor. The customer of this round is the storage company. The garage is the parts counter. The plant comes after the proof.

People already pay for this work

Who already does it The evidence What Marysville takes
Recyclers $32 billion of US sales. About 4 million vehicles a year in the US and Canada. More than 9,000 locations The bay, with a robot on the car
Garages and resellers $12 billion of US used-parts wholesale The parts that come off the car
Storage The electric salvage car already contains the pack The first check, for a graded pack

Recycler sales and the 4 million vehicles are from the Automotive Recyclers Association. Used-parts wholesale is IBISWorld. The behavior is already a business. Marysville puts a robot in that bay.

The same missions, in harder plants

Mission on the car Dexterity Injury in the bay Same mission elsewhere
Unbolt and unclipPack cover, cabin, underbody Clips and blind reach Cuts and pinch Ship engine-room bolts
Pulp-mill pump rebuild
Drop the packAlign and lower the battery A guided heavy lift Crush and high voltage Paper-machine roll change
Ship engine pull
Isolate high voltageProve dead, then disconnect Connectors and interlocks Shock and arc Ship switchboards
Refinery switchgear
Cut it freeSeized bolts, adhesive, shields Force once the fastener fails Sparks and laceration Shipyard pipe cropping
Pulp digester opening
Build it backTorque, connect, and test The teardown, reversed Stored energy on retest Ship diesel back-fit
Paper-machine press rebuild

The car bay is where the motion is learned. The trained robot can be deployed at ships, paper pulp industry, data centers, etc. where that same motion costs the most.

InsightAutomotive Disassembly and Assembly Workflows mimic Industrial Maintenance.

Salvaged cars are the starting point.

A general purpose robot cannot complete these missions

General purpose humanoids

  • Built for free-space reach and a repeated path
  • Actuators saturate easily
  • Very complex control for simple tasks such as opening a bolt
  • High cost of production, not enough value

In case of failure

  • A failed joint, cable, or hand takes the whole robot down
  • The fix is a specialist job and the mechanics have to wait weeks for a fix
  • Legs and hands add parts that the core mission never uses

The bay still needs sufficiently general purpose actuators with a mix of tools

  • Every car, ship, and mill presents a different reach and load
  • Tooling and retooling is needed to disassemble parts needing repair


What has to exist Sensing, an actuator, and control built for contact and for repair on the line

Sensing

  • Force, slip, and release at the fastener
  • Voltage, proximity, and heat before the tool starts work
  • Cameras and RFID to localize accurately
  • Contact sensor for safety around humans

Actuator

  • Drill with electronic torque specification
  • Submodules that unplugs and swaps on the shift with minimal work

Control

  • Move until contact, then limit the force
  • Stop, back out, and retry when the part stays stuck
  • Use LLMs and repair manuals to determine next steps

Hours, then fewer hours

Capture

  • Agentic wearables record a person doing the job on the car
  • The demonstrations come from real salvage cars in the Marysville bay

Reach the last mile

  • Imitation covers the breadth of fasteners and models
  • Safety-constrained learning pushes one task toward reliability, on hardware we own

Compound

  • One model, shared across cells and later across plants
  • The scoreboard is the time to put the next car, then the next plant job, on the robot


The bet, kept narrow Hands and a policy, on hardware we own

The robot

  • Fixed or wheeled, and bimanual
  • Good hands, vision, and force
  • The actuator swaps on the shift

The price

  • Per machine-hour or per part, once the robot leaves our bay
  • Changeover stops being a capital project
  • CDME is the testbed. Marysville is first

The first experiment

  • Will wearable data become a robot policy?
  • That embodiment gap is the seed risk
  • Answer it on our cars before any external plant

The companies a buyer can hire today

On site

  • Path, GrayMatter, Formic, Machina, Vention, Intrinsic
  • Field engineers close the last 15%
  • Revenue scales with headcount

One task

  • An 85% system can ship
  • The next fastener starts from scratch
  • A new car, and a new plant, is a new project

The buyer’s ask

  • A plant needs about 99.9%, a recovery, and a way to diagnose the miss
  • The garage wants the robot back on the car the same shift
  • Hours to the next part, on hardware we own, is the scoreboard

Demand for the work is already there. Each of these deployments is still a project, staffed by field engineers.