Tenon
Close-up of a parallel jaw gripper on a production line

Sim-trained bin picking

Policies trained in simulation, shipped to the cell.

Tenon Forge compiles your cell and your SKU manifest into a USD scene, trains the pick policy on GPU, and deploys it as a TensorRT engine on Jetson Thor.

Request a pilot Datasheet v0.4

91%mean held-out success, 400 SKUs

ISO 13849-1 — PL d documentation suppliedCE 2006/42/EC — by integratorUR10eFANUC M-10iDRobotiq 2F-85Jetson AGX Thor 60 W

Why picking projects fail

The break point is SKU 400, not the robot.

The demo manifest

Vendors validate on twenty SKUs picked to work. Real manifests run 2,000 to 8,000, and the tail is where grippers fail: soft polybags, crushed cartons, nested trays, film under 40 µm.

Every exception is a person

Each failure becomes a manual exception and each exception becomes someone standing at the cell. Picking projects do not fail on kinematics. They fail on the long tail nobody budgeted for.

Product — build 0.4.118

Three modules, one pipeline.

Product and deployment

1. Scenario compiler v0.4

Imports cell CAD and the SKU manifest and emits a USD scene with physically plausible randomization of friction, mass, lighting and depth noise. 4,096 parallel environments per run.

2. Policy trainer v0.4

PPO with imitation warm-start from teleoperation, trained in Isaac Lab on an 8×H100 node. Every generation is archived by seed. 90 GPU-hours per generation.

3. Edge runtime v0.4

TensorRT engine on Jetson AGX Thor at 30 Hz. Falls back to cuRobo replanning when grasp confidence drops, then releases to the exception lane. Runs offline; the plant needs no internet.

Specifications

Measured, not estimated.

Cycle time2.9–3.4 s at 2.1 kg payload, test cell, October 2026
Grasp success91% mean held-out, 400 SKUs, internal benchmark 2026-10
Success definitionLifted, transported and released inside the target bin without regrasp, 2 cm placement tolerance, three attempts maximum.
Throughput1,100 picks per hour net, 2 exceptions per shift, 3.1 s cycle

Full specifications

Training runs on an 8×H100 node in burst regime; 1,980 GPU-hours and US$ 4,400 a month.

Scope

What Forge does not do.

Transparent film under 40 µmDepth sensors lose the surface. 72% held-out, below the 88% gate. With a vacuum array the same family reaches 89% and clears.
Parts under 8 mmJaw geometry is the limit, not the policy.
Moving conveyorsScene must be static during a cycle. Conveyor tracking is roadmap, not shipped.

What it does not do

Pricing

Published, no quote dance.

Pilot

US$ 4,500

One cell, three families, six weeks, gate report

Production

US$ 890 /mo

Per cell per month, annual, offline licence

New SKU family

US$ 2,400

Training plus gate run, waived under 88%

Included: scene compilation, retraining, edge image, upgrades. Not included: arm, gripper, safety hardware, integration labour, lighting.

Lead time: pilot starts within 15 business days of purchase order. Gate pass to handover: 52 minutes.

One engineer on site in weeks 1 and 3; travel and per diem included in the pilot price.

Below 88% on a new family is not invoiced. The family stays in exception routing until it clears.

In production today

24 cells in production with 3 customers.

ARR US$ 256k, first invoice June 2026. 6 more cells on the bench.

Customers by sector and size
Regional 3PL14 cells
National logistics operator6 cells
Parcel distributor4 cells
Carton boxes moving on a roller conveyor

Customers

Pilot result, regional 3PL.

“Dejamos de comprar robots que solo funcionan en la celda de demo del proveedor. El cambio de referencia pasó de tres semanas de tuning a menos de una hora.”

Head of automation

Regional 3PL, 14 cells

3 weeks on-site tuning per changeover, 71% success at SKU 400

Before

Per changeover, on site

41 minutes, 93.4% at SKU 400, 2 exceptions per shift

After

With the first Forge policy

Reference call available on request with the head of automation at the 3PL.

Company

Built in Córdoba.

Founded November 2023.

Jose LopezCo-founder, CEO · 8 yrs industrial automation
Camila FerreyraCo-founder, robotics engineer · grippers, safety validation
Tomás IriarteCo-founder, head of ML · RL, sim-to-real
Lucía BracamonteCo-founder, ML engineer · inference, TensorRT

Open roles are on careers.

Office: Av. Juan B. Justo 4650, Piso 2, X5000 Córdoba, Argentina. Bench cells: Parque Industrial Córdoba, Ruta 19 km 4.

Buyer questions

Asked on every call.

How long for a new SKU family?

Five working days including training and gate run. Under 88% the family is not invoiced and stays in exception routing.

Can we see it before paying?

Send six SKUs. We run a free 48-hour bench test on the Córdoba cell and return the per-family success table.

Send six SKUs. Get the success table in 48 hours.

Bench test on the Córdoba cell, free, no commitment. If your tail is above 88%, we will tell you where it breaks.

Start a bench test WhatsApp +54 9 11 2544-3894