Picominer®  

A Lightweight AI/ML Runtime

AI applications need not be ultra large scale. We aim to keep it simple with rule-engines.

Picominer-v1.6-01.01-Picominer-Introduction
CODING THE BUSINESS CONTEXT

Field Data Services

The majority (some reports say nearly 60%) of AI/ML projects never make it past the PoC (proof of concept) stage, despite the fact that the data models look promising in the lab.

The data lifecycle is very different from the software development lifecycle.
Typically, the cost of deploying AI/ML models is not considered in initial estimates.  

The Model-To-Business phase of the AI/ML project lifecycle requires additional processes to adapt the models to the target environment (application context). It serves as the motivation for Picominer's design.

Picominer-Edge-v1.7-04.00-vExpert-3
SMART DECISIONS

Out Of The Box 

Picominer is an out-of-box solution to extend the capabilities of Edge gateways and Cloud applications. The platform offers wide rage of integrations to connect the filed data to enterprise / customer applications.

undraw_factory_dy0a
Picominer-General-Brief-1.9.2
MOTIVATION

Binding the expert knowledge for the target application context.

SOLUTION
A lightweight semantic layer, and AI/ML model serving software for the edge data.
BENEFITS
Flexible MLOps. Easy integrations with data applications.
SOFTWARE

Salient Features

Connectivity

Data acquisition from field devices with Modbus/TCP, OPC-UA, CAN, MQTT, BACnet, Websockets and many other data sources.

Integrations

Unified data simplifies integrations with a wide range of automation devices, MES / ERP / HMI, RPA, Web and Mobile applications.

Implementation

The Picominer platform designs and reference implementation are open source.There is no vendor lock-in.

ESSENTIAL PROCESS OF

Field Data Management Services

PROJECT STUDY

To establish the scope, we will first schedule meetings
with customer teams.

DATA FLOW DESIGN

Determining numerous data touch points that are in line with business/application goals. 

DATA PREPARATION

Target contextualization
Tagging, labeling and semantic transformation of the field data

MODEL REGISTRY

A model registry is created once models are tested against
the target environment. 

INTEGRATIONS

Integration (API) with MES/ERP/HMI and other
information applications.

Benefits that come with our engagement

Reference designs

Valuable reference design, implementation and support for Edge AI applications.

Flexible

Adapt the Edge AI to the needs of the applications. No opinionated framework.

DevOps

A flexible but uncomplicated DevOps foundation for running and scaling up the Edge AI models.

Open source

We have made all of our reference implementations available as open source. No vendor lock-in.

AI R&D

Access to our ongoing research and development on AI/ML sources and reference designs.

Team training

Get the staff a systematic Edge AI training (online or offline, as needed).

Innovation

Our strategy is to empower the teams to innovate in response to market demands.

Standards based

Continuous and incremental process improvement with industry standards and protocols.