reFAB
Stealth Investor access

reFAB

Physical AI for advanced manufacturing.

We're building the decision engine for high-value metal repair — so the parts that get scrapped today come back into service tomorrow.

Investor brief · Confidential

Sign-off-ready repair decisions for FAA-regulated parts.

A cobot scans the damaged part. The digital twin simulates the repair. Side by side, in real time.

CAM 01 Live
Real cobot · Scan → Repair
RealSense D-series · 6-DOF · ~90 sec
SIM 01 T+0
Digital twin · Simulated repair
Physics-grounded · Per-pass verification
Talk to founders

Sign-off-ready repair decisions for FAA-regulated parts.

Aerospace and defense shops scrap hundreds of millions of dollars in repairable parts every year — not because the physics doesn't work, but because a senior engineer's signature is the bottleneck. reFAB scans the damaged part, decides whether and how to repair it, simulates the result on a digital twin, and outputs a regulator-ready audit package. We start with engine airfoils on commercial aero engines.

Stage Stealth · Pre-seed Vertical Aerospace MRO Stack 3D scan + physics sim + agentic planning Founded 2026

A trillion-dollar asset base, repaired by hand and by memory.

Today, whether a damaged turbine blade, structural casting or rotating component comes back into service depends almost entirely on the judgment of a small number of senior repair engineers. They look at the part. They decide. They sign. The institutional knowledge lives in their heads — and it is walking out the door.

A
Manual triage
Repair-or-scrap is a senior-engineer judgment call made part by part, often offline and undocumented.
B
Knowledge attrition
The experts who can make the call are retiring faster than they can be replaced.
C
Tightening oversight
FAA continued-airworthiness rules and DoD sustainment audits are raising the cost of getting a repair wrong.

The decision layer between the scan and the robot.

reFAB takes a 3D scan of a damaged part and outputs the full repair package: the recommended approach, a simulated repair on the digital twin, the toolpath ready to run on the robot, and an audit-ready report keyed to OEM, FAA and DoD repair standards.

It is not a CAM tool, a vision system, or an inspection platform — it is the adjudication and documentation layer that ties them together, so shops can confidently say yes, this can be repaired, here is how, and here is the evidence.

The stack
Structured-light + photogrammetric scan ingest Geometry-aware deviation models Physics-grounded weld & AM simulation Digital-twin verification Agentic planning loop Robot-vendor-agnostic toolpath output Regulator-format audit report generation

From damaged part to signed repair package — end to end.

A cobot equipped with an Intel RealSense camera scans a mock damaged part. reFAB generates the repair toolpath, simulates the result on the digital twin, and the cobot executes the repair. No human in the loop, no narration — just the pipeline running.

Demo Scan · Plan · Simulate · Repair ~90 sec · raw cut

Hardware: 6-DOF cobot · Intel RealSense D-series depth camera · mock damaged airfoil. Software: reFAB scan-to-toolpath pipeline + digital twin. Video link drops here when the cut is ready.

Three steps. Each one a defensible product surface.

STEP 01
Scan
Damaged part captured via 3D scan and metadata. Geometry is compared against the as-designed and as-served baselines.
STEP 02
Decide & plan
Engine evaluates feasibility, selects a repair strategy, simulates the result on the digital twin, and generates the robot-ready toolpath.
STEP 03
Document
Audit-ready report — assumptions, evidence, simulated outcomes — sized for OEM, FAA and DoD sign-off. This is the moat.

The cost curves crossed.

Workforce
The cohort of senior repair engineers who hold the institutional knowledge is retiring on a 5–10 year clock. There is no backfill.
Compute
Geometry-aware models and physics-grounded simulation are finally good enough — and cheap enough — to run per-part, in production.
Regulation
FAA continued-airworthiness expectations and DoD sustainment audits are pushing repair shops toward fully-documented, traceable workflows. Manual sign-offs no longer scale.
Reshoring
National-security policy is pulling MRO capacity back onshore. New capacity has to be built around software, not tribal knowledge.

Repair is the wedge. Re-fabricate is the platform.

The same decision engine that signs off a repair also signs off a re-fabricate, then a new-part build, and eventually any high-value geometry produced by an automated process. Repair earns us the regulatory primitives, the part-class libraries, and the shop relationships. The rest is distribution.

01
Repair existing parts
Aerospace airfoils & structural castings; sign-off-grade audit packages.
Now
02
Re-fabricate what can't be repaired
Same decision and documentation layer; additive- and subtractive-first build paths for legacy and obsolete parts.
~24 months
03
Net-new high-value parts, with sign-off
A physical-AI layer for advanced manufacturing — any high-value geometry, any regulated workflow.
36+ months

Aerospace MRO first, by part class — not by customer.

We focus on part classes where a single repair decision is worth six figures and a single mistake is worth a grounded fleet. Adjacent markets unlock once the decision engine is validated.

Aerospace MRO Defense sustainment Energy / turbines Industrial OEM Additive (re·fabricate)
OEM-captive repair toolingPratt & Whitney · GE Aerospace · Rolls-Royce
Single-vendor, slow to release, captive to the OEM's parts catalogue. Don't serve the long tail of independent MRO shops.
New-part fabrication startupsHadrian · Machina Labs · Divergent
Build new parts at scale. Different physics, different regulatory surface, no repair-decision layer.
Inspection & CMM vendorsHexagon · Zeiss · Creaform
Measure the part. Don't decide what to do with it — that's a senior engineer's job today.
CAM & robot toolpath softwareMastercam · Siemens NX · Robodk
Execute the plan. Don't generate it from a damage scan, and don't produce sign-off documentation.
reFAB is the decision and documentation layer that sits between scan, simulation and execution — the only piece that ships a regulator-grade sign-off package. We integrate with the rest of the stack; we don't try to replace it.

Built by people who've shipped repair, robotics and regulated AI.

A two-founder team with complementary depth in advanced-manufacturing operations and applied AI — backed by advisors from inside the shops and the regulators we sell into.

Founder name
Co-founder & CEO
2–3 sentence bio. Lead with the most credible role in repair / MRO / aerospace operations. Owns customer, regulatory and GTM. Previously: prior role at prior company.
Founder name
Co-founder & CTO
2–3 sentence bio. Lead with the most credible role in robotics / simulation / applied ML. Owns engine, simulation and toolpath generation. Previously: prior role at prior company.
Advisors
  • Former head of repair engineering, Tier-1 OEM
  • Former FAA continued-airworthiness lead
  • Former COO, independent MRO
  • Robotics / physics-sim academic

We're raising pre-seed and signing design partners.

If you fund deep-tech or physical-AI at the pre-seed stage — or if you run repair operations at an aerospace, defense or energy shop — we'd like to meet. Pick whichever surface works for you.