Wedge control
Layup / AFP-control or cure-control or NDT AI for one cell, factory edge-deployed.
From wedge to system of action
Land narrow, expand relentlessly. Trust before autonomy. A compounding data moat at every step.
4
Horizons
MVP
Shipping now
$0M
Yr 1–3 ARR target
upper range
0
Autonomy levels
Horizons
The plan, in four moves
0–6 months
MVP & design partners
Onboard 3–5 partners on the wedge. Ship connectors, the agent loop, review console and assurance-grade audit log. Prove accuracy and scrap / yield / strength ROI in shadow then assist mode. Establish security baseline (SOC 2 Type I in progress; ITAR-ready).
6–12 months
Paid & graduated autonomy
Convert design partners to paid. Graduated autonomy for low-risk AFP / cure cells and hit-strength-before-cure twin. Add a second module; deepen integrations. SOC 2 Type II; SSO / RBAC; first enterprise logos.
1–3 years
The full loop
Own layup → inspect → cure → NDT → trim / assemble end-to-end. Multi-segment GTM. AFP / autoclave / NDT-OEM and prime / tier-1 channel scale; international expansion. Data-network products as premium tiers. Scale to $38M–$85M ARR.
3–10 years
System of action
Become the autonomous operations layer for the world’s advanced-composite producers — a platform / API ecosystem across aerospace, space, eVTOL, wind, hydrogen, EV and defense.
Shipping now
What the MVP does
Agent loop
A constrained, production-grade sense → reason → act loop with human-in-the-loop checkpoints.
Connectors
AFP / ATL, vision-profilometry, autoclave / OOA, ultrasonic / thermographic NDT and MES.
Assurance log
Grounded outputs and an immutable, assurance-grade audit log from day one.
Autonomy ladder
How control expands
Shadow
Observe and predict, zero control authority.
Assist
Recommend and predict; engineer approves.
Bounded auto
Autonomous inside qualified envelopes; escalate exceptions.
Fleet
Multi-cell, multi-site autonomy with edge fleet management.
Data products
The network raises the floor
As the corpus grows, Curelum ships data-network products — porosity, strength, scrap and cure-energy benchmarks — as premium tiers, and a federated validation network that shares rare defect signatures without exposing recipes or ITAR-sensitive data.
- Porosity / strength / scrap / energy benchmarks
- Federated, IP-preserving validation network
- Cross-material and cross-geometry generalization
Guiding principles
What won’t change
Trust before autonomy
Grounding, audit, human-in-the-loop, graduated control.
Compounding moat
Every supervised correction trains the system.
Land narrow, expand
Win the wedge; become the system of action.
Metrics
How we’ll know it’s working
0
Design partners
year one
0%
Target NRR
0pts
Yield gain
SOC 2
Type II
year one
FAQ
About the roadmap
How fixed is this plan?
The horizons and principles are firm; the sequencing flexes with design-partner feedback. Partners help decide what ships next.
When does autonomy actually control machines?
Only after shadow and assist prove accuracy and ROI, and only inside qualified envelopes on low-risk cells first.
What unlocks the data-network products?
Scale — enough supervised outcomes across sites to make benchmarks and a federated validation network valuable.
The endgame
Every flight-critical part placed, cured and certified on Curelum.
Shape the roadmap with us
Design partners help decide what we build next.