A pharma plant can't be disturbed to be understood. Elpis collects from it read-only — beside your existing control systems.
Batch lines, packaging and filling machines, cleanrooms, and the critical utilities that keep them running — HVAC, compressed air, purified and WFI water, chillers. Most of it is PLC-controlled (Siemens S7, Modbus). Elpis reads every controller as a read-only client, normalizes the signals to one vocabulary, and watches the rotating utility equipment that fails quietly — without changing any control logic. Operating across India and the Middle East.
Live integrations: FOCAS2 · MTConnect · Brother HTTP · Modbus TCP · OPC UA Client · Siemens S7. FANUC MT-LINKi REST on the roadmap.
The control systems stay untouched. The data layer is on its own.
A pharmaceutical plant is built around equipment that cannot simply be touched. Batch reactors, filling lines, blister and cartoning machines, the cleanrooms they sit in — each was commissioned for the product it runs and locked under change control. Most of it is PLC-controlled: Siemens S7 on the process and packaging cells, Modbus on the older skids and balance-of-plant. None of it was installed to be queried by a monitoring platform, and none of it can come offline just to be connected.
Around that change-controlled core runs an entire plant of critical utilities — HVAC holding cleanroom pressure cascades, compressed-air systems, purified-water and WFI loops, chillers, and the pumps, compressors, and fans that drive them. These are rotating machines, and they fail the way rotating machines do: quietly, then all at once. A utility excursion doesn't just stop a line — in a regulated context it can put a batch in question. Yet the early signs are usually on the floor long before anyone acts on them, scattered across PLCs and standalone sensors that never reach a trend.
The answer is not to re-touch change-controlled equipment. It is to modernize the data layer beside it. The pharma plants that get there put one protocol-agnostic runtime in front of every controller as a read-only client, normalize every signal at the edge, instrument the consequence-heavy utility equipment, and let the existing control systems keep doing exactly what they were commissioned to do — unchanged.
"The early signs are on the floor long before anyone acts on them."
Read-only, beside your existing control systems.
Read-only, beside your existing control systems.
The outcomes a pharma plant reaches for.
Built to sit beside your control systems, not inside them.
Elpis is deployed across regulated and process-driven manufacturing operations — plants with change-controlled equipment, PLC-controlled lines, and critical utilities. Operating across India and the Middle East. EdgeConnect connects as a read-only client and sits beside your control systems: it changes no control logic and brings no machine offline. The platform runs offline-first — the license validates locally with no phone-home, and per-route store-and-forward is built to preserve every reading through a network or broker drop, queuing locally and replaying in source order on reconnect. Every configuration change to EdgeConnect itself is captured in a hash-chained, tamper-evident change record that supports your existing change-control discipline.
Full operational trust posture → /security · Anonymized deployment patterns → /customers
What pharma teams ask first.
Which of our equipment can you actually collect from?
Siemens controllers over S7, older skids and balance-of-plant over Modbus TCP, plus FOCAS2, MTConnect, Brother HTTP, and OPC UA Client where it's exposed — all shipping today. That covers most PLC-controlled batch and packaging lines and the utility equipment around them. FANUC MT-LINKi REST integration is on the roadmap. Bring the equipment list to the scoping call and we confirm the collection path per machine.
Does this touch or change our existing control systems?
EdgeConnect connects to each controller as a read-only client — it reads data and changes no control logic, and no machine comes offline to connect it. It sits beside your existing control systems rather than inside them; the control side keeps running exactly as it does today, and Elpis only observes. How that read-only data path fits your site's own change-control and quality processes is your determination to make — we're glad to walk your engineering and quality teams through the exact data path on an architecture review so they can assess it against your procedures.
Can you monitor our utilities and rotating equipment for early failure?
Yes — that's what VAS and mDAQ are for. VAS reads vibration signatures on rotating utility equipment (HVAC fans, chillers, compressors, process pumps); where a skid exposes nothing useful, mDAQ acquires temperature, pressure, flow, or vibration directly. Both give early warning when a signature crosses a threshold your maintenance team defines — a better trigger than a calendar, not a guarantee against every failure.
How does the change record help our change-control discipline?
Every configuration change to EdgeConnect is captured in a hash-chained, tamper-evident change record — what changed, and that the history hasn't been altered after the fact. It is designed to support the change-control discipline you already run, giving you a verifiable record of changes to the data layer. It is a change record, not a compliance or validation deliverable — your own change-control processes and your quality system remain the system of record. → /security
Does this replace our SCADA, MES, or historian?
No. Elpis sits beside them. EdgeConnect publishes canonical signals read-only (MQTT, OPC UA Server); EREMOS V2 exposes OEE, alarms, and reports via API. Your SCADA keeps operator HMIs and control, your MES keeps batch and scheduling, your historian keeps its records. Elpis adds an operational and condition-monitoring view alongside them. → /architecture
Looking for the same thing from another angle?
Bring us your plant — change-controlled equipment and all.
An equipment and utility list, the rotating equipment that worries you, and an OEE definition — that's enough to scope a proof of value. We run it read-only, on your real protocols, beside your real systems.