Industrial software for oil & gas and EPC.
Plant software, proven on your own data. We are not a generic software house that happens to have an industrial client. Every system here was built against real plant data, real protocols and real operating procedures — and handed over as something an operating team runs every shift.
Different problems. One delivery stack.
Construction execution, engineering design automation, maintenance and reliability, protocol integration and closed-loop optimisation — each one delivered on a real plant, each one built on the same five layers.
Progress Control
Daily construction progress, manpower, blockers, QA/QC and HSE with KPI accountability — one auditable record from weekly target to client view.
ReviewMate
AI review of P&IDs and engineering documents against your company standards — every comment clause-cited and verified before it is raised.
P&ID Studio
Tools for engineers who work with P&IDs, built on the DEXPI standard: convert drawings, check them, pull registers, compare revisions, and prepare isolations and HAZOP worksheets.
AssetOps CMMS
Asset register, work-order lifecycle, preventive maintenance, spares and SAP procurement, with an MTBF / MTTR / availability KPI engine.
Plant OTS
A 3D operator training simulator of the whole plant — live process model, instructor fault injection and an interactive equipment library — built from your own drawings and IO list.
Procedural Automation
Operating procedures executed as structured, auditable sequences — state-based event management with automatic execution.
ComGate
Protocol gateway, simulator and test bench for Modbus TCP / UDP / RTU and OPC UA / DA — engineering, polling, traffic inspection and historian in one tool.
Plant Autopilot
Five modules — Parser, VFM, AI/ML, APC and Prioritizer — from tag acquisition through soft sensing and advanced control to prioritised operator action.
Soft sensing & analytics
Virtual sensors, predictive maintenance, anomaly and fault detection, digital twin and real-time optimisation — on the same delivery stack.
Projects reuse layers rather than restart them.
Whatever the application, the same five layers appear. Pick a product to see which parts of the stack it stands on — and what already exists to build your scope on.
Numbers from the test reports, not the pitch.
Every conversion produces its own evidence pack; every protocol is exercised against third-party tools; every forecast states its error. These come from the qualification records.
Bound endpoints on a reference P&ID — 428 components, 1,021 connections.
Imported as ComGate tags; 367 writable, write path proven.
On the Autopilot forecasting module, with MAE · RMSE · MAPE reported.
Construction blockers owned, escalated and closed — on a governed project record.
Your drawings. Your wells. Your plant.
A scoped proof of concept on your own drawings and tags, an accuracy gate agreed up front, evidence produced against it — and a decision to scale taken on measured results.
Scoped on your own drawings, wells or units — with the accuracy gate agreed before we start.
Automated test suites per project, regression gates in CI, thresholds that fail the build when they slip.
Signed Windows installers with embedded runtimes and supervised services — or hosted on plant infrastructure.
Role-based access, audit logging, operating manuals and architecture notes handed over with the source.
See a review end to end.
ReviewMate, our engineering document review platform, in three and a half minutes — from the approved requirement register to a signed, clause-cited comment, on an illustrative drawing set. More on ReviewMate →
“SMHcoders delivered exactly what they scoped — clear communication, clean handover, and a model that performs in production, not just in the demo.”
Latest field notes.
Industrial AI27 Jan 2026 · 7 minWorkflow automation in industrial operations: closing the loop from alarm to actionMost unplanned downtime is not lost while the machine is failing. It is lost in the hand-offs around it: the alert nobody acknowledged, the diagnosis waiting…Start with a proof.
Happy to walk through any of these systems in detail, or to scope a proof of concept against your own drawings, tags or historian data.