From Integrators to IntegratorsFrom Integrators to Integrators

Coreflux is the programmable MQTT Broker that enables a full data management platform. Connectors bring in data from machines, devices and business systems, and LoT, the Language of Things, structures that data, runs the logic and sets who can access it, all inside the Broker. The Coreflux Hub shows everything live, from single values to full Dashboards. It runs on edge devices, on your own servers or in the cloud, and a network of system integrator partners delivers it.

Most MQTT brokers move messages between clients. Coreflux is a programmable MQTT Broker: Connectors, data Models, event logic and access Rules all run inside it, written in LoT. There's no separate gateway or rules engine to keep in sync, and the whole system becomes one readable description that people and AI can both work from. The next site starts as a copy of the last one.

LoT, the Language of Things, is the language Coreflux uses to define what happens to real-time data inside the MQTT Broker. It reads close to plain English and has four constructs: Connectors link machines and systems, Models give data a shared structure, Actions run logic when a value changes or on a schedule, and Rules set who can read or publish each topic. LoT is the SQL of real-time MQTT data, and both people and AI assistants can write and review it.

Coreflux connects PLCs and other equipment with Connectors that run inside the MQTT Broker, so no separate gateway software is needed. Industrial Connectors cover Modbus TCP and RTU, Siemens S7 (S7-300 to S7-1500), Allen-Bradley (ControlLogix and CompactLogix), EtherNet/IP, OPC UA, Beckhoff ADS, Omron FINS, BACnet/IP and Sparkplug B. Pick the protocol, fill in a short form or a few lines of LoT, and each data point lands in its place in your namespace. OPC UA Server and Modbus Server Connectors also let an existing SCADA read the same data.

Coreflux writes MQTT data to databases with Connectors that run inside the Broker: PostgreSQL, MySQL, MariaDB, SQL Server, MongoDB, OpenSearch, CrateDB and file storage. The same Connectors send data to Snowflake, Kafka, REST APIs, email and other MQTT brokers. LoT Models shape each reading and an Action decides what gets written and when, so data reaches the database already structured and named.

Coreflux gives AI agents structured real-time data with clear permissions, because AI can't safely act on a system it can't read. Models give every value a name, unit and context, and Rules decide what an agent may read or publish. Actions can call MCP servers and AI agents directly. AI tools can also connect to the Broker's own MCP server, where every call runs with that user's permissions, and Coreflux's documentation MCP server gives them accurate Coreflux syntax. In the Hub, the AI Assistant runs on your cloud LLM, writes changes as ordinary LoT, and deploys nothing until a person presses Deploy.

Yes. A Unified Namespace (UNS) is one organized, real-time view of an operation's data, often structured by ISA-95 levels: enterprise, site, area, line and machine. In Coreflux, the MQTT Broker carries the data, LoT Models give every source the same structure, and Rules control access to each branch. Sparkplug B Connectors join existing Sparkplug systems, and the namespace extends up to KPIs and business systems. The Hub's UNS Viewer shows which logic publishes and reads every value.

In Coreflux, you build MQTT dashboards in the Hub: Dashboards bind gauges, charts and controls to live MQTT topics with no code. You can also describe the dashboard you need and let the AI Assistant build it. Every Dashboard reads the same data as the rest of the platform, so any value leads straight to the logic behind it. Share one as a read-only link, or run it full screen on a control-room display.

Coreflux runs at the edge, on your own servers or in the cloud. The MQTT Broker installs on Windows, Linux and Raspberry Pi, runs in Docker or on Kubernetes, and is also offered as a managed broker in Coreflux's cloud. Brokers at different sites bridge into one system, and clustering adds capacity and failover. The same architecture that starts on a Raspberry Pi scales to enterprise.

Yes. LoT Actions can call Python functions with CALL PYTHON for work that's easier in a full programming language, such as statistics, machine learning models, advanced text parsing, or libraries like NumPy and pandas. LoT still decides when the Python runs, and it runs inside the Broker, so there's no extra service to deploy.