Faber retro-modern landing — bright cream surface, contemporary typography, electric blue + violet accents.

Product
Thesis
Industries
Partners
Get access
Forward-deployed engineering · physical AI

The AI-native systems design platform for warehouse automation.

Faber is being built for system integrators and robotics OEMs deploying physical AI today. Engineers and AI agents work side by side to design, validate, deploy, and operate automation systems — and the engineering work compounds across projects, not just within them. Starting with warehouses.

faber · atlas-site-04 Preview · concept tour
01Concept
02Design
03Simulate
04BOM
05Deploy
06Operate
Project
requirements
layout
throughput
fleet
bom
schedule
operate
Status
v0.1 scope
v0.2 design
v0.4 current
v0.5 next
Requirements · 7 captured · 96 × 34 m site
REQFIELDVALUESTATUS 01throughput4,500 picks/hr● confirmed 02floor area96.0 × 34.0 m● confirmed 03ceiling11.5 m● confirmed 04hours / day16● confirmed 05peak factor1.4 ×○ pending 06sla / charge≥ 10%○ pending 07budget$ 4.2M cap○ pending site sketch
96.0 m 34.0 m A · STORAGE — 8 aisles B · PICK + PACK C · CHARGE
picks / hr 5,0004,0003,0002,0001,000 06:0010:0014:0018:00 target 4,500 peak 4,820/hr
#COMPONENTVENDORQTYUNITTOTAL 01AMR-12 robotvendor-A12$42K$504,000 02rack 11.5mvendor-B192$850$163,200 03charger 50kWvendor-C8$18K$144,000 04conveyorvendor-B240 m$420$100,800 05floor markersvendor-D240$45$10,800 06WMS licensevendor-E1$185K$185,000 07installvendor-A1 lot$420K$420,000 + 77 more$2,652,200 TOTAL · 84 line items $4,180,000
W1W3W5W7W9W11W13W14 P1 site prep P2 floor mark P3 racks P4 chargers P5 robots P6 integ. P7 accept. — legend — critical path non-critical
PICKS / HR 4,820 ▲ +14% vs target FLEET ACTIVE 12/12 all online UTILIZATION 0.92 peak shift CHARGE BUFFER 14% ▲ above SLA LAST 24h THROUGHPUT
Agentactive
Concept agent
Captured 7 reqs. Throughput target 4,500/hr. Confirm peak shift?
Layout agent
Buffer drops to 6% — below SLA. Recommend +2 chargers in zone C.
Sim agent
Simulation complete. Sustained 4,820/hr. Bottleneck: zone B aisle 4.
BOM agent
Auto-generated 84 line items. Total $4.18M. Vendor coverage 100%.
Schedule agent
Critical path 84d. Chargers → robots → integration. 14 weeks.
Operate agent
Site live. 4,820/hr · buffer 14%. No anomalies.
0/7 reqs validated · est. footprint 3,264 m²autosaved 3s ago
192 racks · 8 robots · 6 chargersautosaved 12s ago
Sim · 10K cycles · 4,820 picks/hr · util 0.92completed 4m ago
84 line items · $4.18M · vendor coverage 100%auto-gen 18s ago
14 weeks · critical path 84d · 7 phasesscheduled 1m ago
Live · 4,820/hr · NPV $4.21M · util 0.92last sync 12s ago
Stage 01 · Concept
One model · every stage

From the first requirement to the running fleet.

Faber moves with your engineers from the first requirement to the running fleet. Every decision, formula, and assumption lives within connected models — so the knowledge compounds across projects.

01 · CONCEPT
Concept
capture goals + constraints
02 · DESIGN
Design
layout + tech + fleet
03 · SIMULATE
Simulate
throughput + edge cases
04 · BOM
BOM
auto-gen from selections
05 · DEPLOY
Deploy
schedule + commission
06 · OPERATE
4.8K.9241814%
Operate
live models that learn

Accelerate the adoption of physical AI in real-world environments.

Faber is looking for design partners — system integrators and robotics OEMs engineering the platform alongside the teams shipping physical AI today. We want to understand your work and shape Faber around it.

Thesis · what we're building toward

The hard part of scaling physical AI isn't only the robots — it's everything around them.

The friction lives in the seams — between systems, between multi-vendor integration, between design intent and operational reality. Five themes recur across the research, in conversations with operators and integrators, and in the engineering work itself.

01 · integration ownership

The work of holding a deployment together is itself a product.

Across the systems that actually run a warehouse — WMS, WCS, WES, the robotics fleet, the WES-to-fleet connectors — most engineering teams stitch together vendor-specific tools, custom code, and tribal knowledge to make them work as one site. Faber starts upstream: a vendor-agnostic engineering layer that holds the design and the deliverables in one place. We're working toward extending that layer into orchestration — owning the logic that sits across the systems, not just the design that produced them.

02 · connected models

One model that survives from concept to commissioning.

Today the layout, the throughput estimate, the fleet size, the BOM, and the financial model live as separate artifacts in separate tools — maintained by separate people, reconciled by hand. By the time a customer asks "what changes if we drop the budget by 15%?", the model that won the deal is too far from the model that's being built. Faber connects them, so what-if scenarios and impacts are visible immediately instead of six months later.

03 · deployment gap

Closing the gap between what's drawn and what gets installed.

A finished design isn't a deployable site. Survey packages, layout variants for site realities, commissioning deliverables — the work between "design approved" and "robots running" is where weeks disappear, today scattered across PDFs, emails, and one-off engineering. Faber is being built to produce these deliverables from the same model that holds the design, so the path from CAD to physical world is engineered, not improvised.

04 · post-deployment adaptability

The model that wins the deal shouldn't go stale at go-live.

The simulation that closed the project becomes a slide deck the day commissioning starts. SKUs shift, volumes diverge from forecast, peak shifts find bottlenecks the design assumed away. Faber keeps the engineering model live, fed by what the site is actually doing — so the engineers responsible for the project can re-optimize continuously, with the same model they used to design it.

05 · application-engineering scale

Every deployment shouldn't start from scratch.

Each new customer means re-translating their requirements, re-configuring fleets for their building, re-integrating with their WMS. None of this engineering work compounds — the model from the last project doesn't make the next one faster. Faber is being built so it does. What's learned on one deployment becomes the starting point for the next.

Where we are headed

As agents take on more of the engineering work, the human role shifts from doing to deciding. Faber is being built as a forward-deployed engineering platform — where engineers and agents collaborate on modeling, translation, and configuration, and engineers make the judgments that matter.

Roadmap

Starting with warehouses. Building toward everywhere physical AI gets deployed.

Faber starts with warehouse automation — the segment we know best, and where the latest wave of robotics is being deployed at scale. The roadmap extends outward into manufacturing, factories, and data centers, wherever physical AI gets deployed.

Now

Warehouses

Full lifecycle — concept through operations. MFC, fulfillment, dark stores.

Active build
Exploring

Manufacturing

Cells, lines, inline QC, robotic work-cells.

2026+Roadmap
Roadmap

Factories

Multi-process facilities, brownfield retrofits, greenfield builds.

2027+Long-term
Roadmap

Data centers

Robotic provisioning, cooling systems, physical infrastructure.

FutureLong-term

Have a domain we should build for next?

If you operate physical infrastructure that doesn't fit the categories above — let us know. We want to understand where the engineering work of deploying physical AI is hardest.

Design partners

Build the platform alongside us.

Faber is being built for the engineers shipping physical AI today. We're partnering with a small number of system integrators, robotics OEMs, and operator-integrators to shape the platform around the work you actually do.

01

System integrators

Teams designing and delivering automation projects. Faber is being built around the workflow we've seen at integrators — from first customer conversation through commissioning.

02

Robotics OEMs

Teams selling robotic systems into integrators and operators. Faber is being built so the application engineering on every new customer compounds across deployments instead of starting over.

03

Operator-integrators

Internal automation teams running brownfield and greenfield projects in-house. Faber is being built so engineering rigor scales across projects without scaling team size.

Become a design partner

Tell us about your work and what you're trying to ship. We'll be in touch.

Faber · 2026