We are factory people who know AI
not the other way around.
Almetra builds AI systems that learn from real shopfloor behaviour — turning factory video into structured process understanding, operational insight, and automation readiness.
The technical core is not classic computer vision. It is long-horizon activity understanding, process mining from video, typed process graphs, large temporal-aware teacher models, teacher-student distillation, and edge-deployed intelligence in real factories.
THE TECHNICAL PIPELINE
From pixels to process graph
six transformation layers.
Manufacturing work unfolds over minutes. Almetra's systems model this temporal structure and convert it into a ground-truth representation of what is happening on the shopfloor.
LAYER 01
Pixels
Raw video
LAYER 02
Entities
People · tools · parts
LAYER 03
Actions
Steps · motions · waits
LAYER 04
Process Events
Cycles · bottlenecks
LAYER 05
Process Graph
Typed · auditable
LAYER 06
Insight & Action
CI · automation
Vision AI
Detection
Temporal model
Event extraction
Graph construction
Improvement & robotics
TECHNICAL STACK
Six transformation layers
from raw video to operational decisions.
LAYER
WHAT IT DOES
TECHNICAL CHALLENGES
The problems
our team works on.
LONG-HORIZON VIDEO
Modelling work over minutes, not frames
Manufacturing processes unfold over minutes with temporal dependencies, action segmentation, and process state that frame-level detection cannot capture.
TEACHER-STUDENT DISTILLATION
Compressing large models for edge deployment
Transferring the capabilities of large temporal teacher models into compact student models that run close to the production line — under real factory latency and bandwidth constraints.
VIDEO-TO-PROCESS GRAPHS
Reliable process events from noisy shopfloor video
Converting raw factory video into typed process graphs with auditable transitions, dependencies, and failure modes — structured enough for operational reasoning and automation decisions.
CROSS-FACTORY ADAPTATION
Adapting across factories with limited labels
Generalising across factories, workstations, operators, camera viewpoints, and product variants without requiring exhaustive annotation at each new deployment.
CYCLE MATERIALISATION
Deriving cycles from dense workstep streams
Extracting clean cycles and per-product bookkeeping from continuous, dense workstep streams — replacing brittle counting heuristics with reliable structured event data.
AUTOMATION READINESS
Which tasks are worth automating — and when
Determining which human tasks are economically meaningful, technically feasible, and operationally safe to automate — grounded in real process data, not lab conditions.
INFRASTRUCTURE & RESEARCH
Credibility grounded
in systems, research, and production deployment.
TRAINING INFRASTRUCTURE
NVIDIA B300 Blackwell-class accelerators
Self-hosted high-performance training infrastructure for training and adapting large temporal models on factory video at scale.
ACADEMIC RESEARCH
Prof. Jürgen Gall, University of Bonn · Prof. Rudolph Lioutikov, KIT
Academic advisors and collaborators in computer vision and robotics — bringing frontier research depth to production deployment challenges.
ROBOTICS HARDWARE
Universal Robots · Franka arms · GELLO teleoperation
Industrial automation work on UR platforms and manipulation research on Franka arms with GELLO-style teleoperation for skill collection and task learning.
DATA ADVANTAGE
Thousands of hours of real factory video at scale
Almetra observes real manufacturing work across hundreds of deployed processes — creating a rare dataset for temporal video understanding, process graph construction, and automation-readiness analysis.
System status · OperationaL
v 2026.5
Watch recent customer story