ANOMX PLATFORM
An AI layer above your data, models, and machines. Anomx connects continuous analysis with an agent that investigates change and prepares what comes next.
EVERYTHING IS CONNECTED
Connect data to the systems that produce it. Channels, datasets, analysis jobs, and findings become understandable and traceable alongside your infrastructure.
INSIDE THE PLATFORM
Create your own ongoing instructions. Your background agent returns to them on a schedule — even when no one has the platform open.
“Infer the system structure and its components from all the data available to you.”
An ongoing instruction turns names, descriptions and connected sources into a system hierarchy that can be reviewed and refined.
“Keep data channels coherent, with consistent units and meaningful descriptions.”
Revisit channel metadata, identify gaps and prepare corrections. Apply changes within the object permissions you define.
“Look for correlations between channels and update the data graph accordingly.”
Compare related observations and propose new graph connections. Keep the evidence attached: correlation is a clue, not proof of causation.
“Train forecasting models for the data channels our team views most often.”
Use available usage context to prioritize channels, prepare training jobs and retain candidate models for evaluation.
Illustrative assignments. Connected tools, data access, permissions and compute determine execution.
BACKGROUND INTELLIGENCE
The background agent revisits connected data and system state on a schedule. It brings emerging deviations into context.
Cooling circuit · time-series analysis
Investigate data, jobs, and related assets together. Use specialist tools and model outputs to turn a deviation into a grounded explanation.
Temperature · flow rate · previous runs
Create a recommendation for review, or make the platform changes your team has explicitly enabled. Permissions and usage budgets define the scope.
Evidence attached · awaiting human review
Inspect earlier background runs, carry forward useful context, and avoid repeating the same findings. A continuous thread of operational intelligence.
Run history · evidence · follow-up
BUILT AROUND REAL SYSTEMS
From research facilities to industrial machines and distributed infrastructure, Anomx brings AI to the places where decisions matter.
Your team chooses which platform objects background agents may create, update, or delete. Other changes become recommendations.
Hourly and daily token budgets govern background work. Reaching a limit pauses execution.
Inspectable runs, evidence, object versions, and human feedback keep the system accountable.
Distributed workers handle data acquisition and CPU or GPU workloads. The platform connects execution with context.
SCIENCE AT THE CORE
Three complementary ways to detect the unexpected. One foundation for informed decisions.
Learn temporal behavior and compare new observations with a forecast. Residuals reveal where reality departs from expectation.
Rolling-window models · Darts integrationscore = | observed − predicted |Illustrative visualization · not measured data
Compress and reconstruct the system’s observations. Patterns that cannot be reconstructed well become candidates for investigation.
Principal component analysis · PyTorch autoencodersscore = ‖ observed − reconstructed ‖Illustrative visualization · not measured data
Compare observations in a feature space. Isolation and normality models surface unusual combinations that single-channel thresholds can miss.
Feature-space modeling · Isolation Forestscore = unusualness in feature spaceIllustrative visualization · not measured data
THE NEXT ERA
Build the future of autonomous systems with us.
Request early access