01 / CONNECT
Plug into the floor
Modbus (TCP + RTU serial), OPC-UA, EtherNet/IP, MQTT, FOCAS, MTConnect. Native adapters. No drivers, no plugin chain.
ModbusModbus RTUOPC-UAEtherNet/IPFOCASMQTT
Know what your machines
are saying.
OpexFlow connects MModbus, OPCOPC-UA, IPEtherNet/IP, 
Self-hosted · single binary · ~38 MB · ARM64 to rack server
flow: machine-telemetry
Machines in · transform in flight · clean data out
From connection to action
Connect anything. Ship clean data.
Four steps from a raw machine signal to a normalized record in your data lake. No glue scripts, no fragile plugin chains.
02 / PULL
One clean stream
Polls every source, dedupes the noise, normalizes on the fly. A compiled async core, not a single thread.
03 / TRANSFORM
Just describe it
Say what you want in plain English. The AI writes the JS or native DSL. You never touch the syntax.
"Convert °C → °F"set temp_f = temp_c * 9/5 + 32
04 / DELIVER
Ship anywhere
ClickHouse, Postgres, NATS, HTTP. HMAC-signed and SQLite-retained, so it survives restarts and gaps.
ClickHousePostgresNATS
✦ AI-powered transforms
Describe it. The AI writes the code.
No hand-written transforms. Tell OpexFlow what comes in and what you want out. It emits JavaScript or the native DSL and drops it straight into the pipeline. Sentence to running code in seconds.
You describe
INPUT
{ machine.id, metadata.temp_c }“Convert metadata.temp_c to Fahrenheit using the native DSL.”
OUTPUT
{ metadata.temp_f }DSL · °C → °FJS · filter anomaliesJS · redact PIIJS · map vendor fields
✦ OpexFlow AItemp.dsl
✦
Plain English → code. Describe the change; the AI emits JS or native DSL. You never write syntax.
✦
Picks the fast path. Native DSL skips JSON marshalling, giving the lowest latency and highest throughput.
✦
Sandboxed & editable. Memory-capped, deadline-interrupted. Always yours to tweak.
Why teams use OpexFlow
Your machines are already producing signal.
OpexFlow turns raw protocol noise into the actions that move uptime, quality, and throughput.
Catch failures before they stop the line.
Vibration and temperature anomalies stream in real time, so you can trigger work orders before a breakdown, not after.
Know your real uptime.
Availability, performance, and quality per machine, as one normalized stream from every protocol on the floor.
Stop defects at the source.
Spindle load and tool-wear signals flag process drift before scrap piles up on the next shift.
Find the bottleneck.
Per-cycle part counts across every line surface exactly where production stalls, in real time.
An audit-ready data trail.
HMAC-signed, SQLite-retained records survive restarts and network gaps. Defensible by default.
One pipe to every system.
Feed CMMS, MES, and your data lake from one engine: NATS, MQTT, ClickHouse, Postgres, or HTTP.
Licensing
Priced like software, not like tags.
Per-node and per-site licensing, never metered by the tag. Transparent tiers benchmarked against what the market actually pays for gateways today.
Starter
Starter
Single edge node for pilots and small lines.
$499 / node
per year · billed annually
- ✓All 8 protocol adapters
- ✓Multi-flow canvas + JS transforms
- ✓SQLite retention & HMAC push
- ✓1 edge node · ARM64 supported
- ✓Email support
Most popularPro
Pro
Production multi-flow deployments for a plant.
$1,490 / node
per year · billed annually
- ✓Everything in Starter
- ✓High availability & clustering
- ✓RBAC + audit log
- ✓OTel traces & metrics
- ✓Priority support · SLA response
Enterprise
Enterprise
Unlimited sites & nodes. One company, one license.
from $19,000 / yr
flat annual · unlimited sites & nodes
- ✓Everything in Pro
- ✓Unlimited sites & nodes
- ✓Air-gapped / on-prem deploy
- ✓Custom adapter development
- ✓Dedicated engineer · uptime SLA
What counts as a “node”?
A node is one running OpexFlow instance: one deployed binary on one host. That host can be an edge gateway, a Raspberry Pi, a VM, or a rack server. One node license covers unlimited flows, unlimited tags, and unlimited protocol connections on that single instance.
- ✓ 1 node = 1 host running the OpexFlow binary
- ✓ Unlimited flows, tags & device connections per node
Pricing benchmarked against published Kepware (~$1,100 platform + ~$452/yr) and Ignition (~$2,000) market data. No per-tag metering. No surprise renewals.
Enterprise license: the fine print, up front
The flat Enterprise rate covers unlimited sites and unlimited nodes, subject to these binding terms:
- ✓Single legal entity. Licensed to one named company / corporate group. Affiliates and subsidiaries under common ownership are included.
- ✓Internal use only. Deploy within your own operations and OT environment: plants, sites, edge gateways you own or operate.
- ✕No resale or re-hosting. Not for redistribution, white-labeling, or offering OpexFlow as a service to third parties.
- ✕No multi-tenant SaaS. Each external customer or tenant requires its own license.
- ✓Non-transferable. Bound to the licensed entity; cannot be sold, leased, or shared with unaffiliated organizations.
Need to host OpexFlow for your own customers? Ask about our OEM / SaaS partner terms.
FAQ
Know before you connect.
Built for teams who want clean machine data without living inside search tabs, dashboards, and glue scripts all day.
01Is OpexFlow a Kepware alternative?+
Yes. OpexFlow covers the same industrial connectivity ground (Modbus TCP, OPC-UA, Fanuc FOCAS/CNC, and MTConnect), but adds flow-based transformation and a canvas, native OpenTelemetry, and a single binary. FOCAS loads Fanuc's fwlib30.dll directly and is Windows-only (the adapter self-disables with a clear status on Linux/ARM); MTConnect gives you cross-platform CNC on any OS. Priced per node or per site, never per tag.
02Is OpexFlow a Node-RED alternative?+
Yes. OpexFlow gives you the same drag-and-drop flow experience and inline JavaScript transforms, but on a compiled, fully async core with industrial protocols built in: no Node.js runtime, no fragile plugin chain, no per-flow memory blowups. One binary from edge to server.
03How does the transform layer work?+
Two engines. JavaScript transforms run in an embedded, sandboxed runtime, deterministic and isolated from the host, with hard memory and CPU limits. The native OpexFlow DSL is evaluated directly against the in-memory event: no JSON marshalling, no interpreter VM, so it's the fastest path for simple maps, filters, and field math. The AI generates either; you never write the syntax yourself.
04Can it run on the edge, on a Raspberry Pi?+
Yes. OpexFlow cross-compiles to ARM64 and ships as a single self-contained binary with an embedded UI. The same artifact runs on a Pi, an industrial gateway, or a rack server.
05What does licensing look like?+
OpexFlow is proprietary, commercially licensed software. Tiers run from Starter (single node) through Pro (production HA) to Enterprise (site license + SLA). No free tier; every deployment is supported and licensed.
06What exactly counts as a “node”?+
A node is one running OpexFlow instance: one deployed binary on one host (an edge gateway, a Raspberry Pi, a VM, or a server). One node license covers unlimited flows, unlimited tags, and unlimited device/protocol connections on that single host. A multi-host high-availability cluster needs one license per node in the cluster.
07Does it integrate with the OpexMX CMMS?+
OpexFlow is the data backbone for the Opex suite. It feeds normalized machine data into OpexMX and any downstream system via NATS, MQTT, ClickHouse, Postgres, or HTTP. It also stands alone as a general industrial middleware.
08How can Enterprise be unlimited sites and nodes?+
The Enterprise rate is a flat annual license for unlimited sites and nodes, but it is bound to one named legal entity (your company and its owned affiliates). It covers internal operational use across every site you own or operate. It does not permit resale, white-labeling, re-hosting, or offering OpexFlow as a multi-tenant service to outside customers; each external customer needs its own license. OEM and SaaS partner terms are available separately.
flow.opexcg.com
Your machines are talking. Start listening.
Connect anything. Transform in flight. One licensed engine. Request access for your floor.