From Clipboard to Cloud: Digitising Precision Engineering in Oil & Gas

In critical bolting applications — flanges on subsea pipelines, structural connections on offshore platforms, pressure vessels in refineries — the difference between a correctly tensioned bolt and an under-tensioned one can be catastrophic. Bolting engineers know this. The challenge is that the documentation systems designed to track it have not changed meaningfully since the 1990s.

The typical field workflow: a bolting technician tightens a joint, records the torque reading on a paper traveller card clipped to a clipboard, and hands the card to a supervisor who eventually files it in a manila folder. That folder might live in a site office, a shipping container, or the back of a pickup truck. Retrieving a specific reading from six months ago to investigate an anomaly takes hours. Sometimes the paper is simply gone.

A major bolt manufacturer serving the petrochemical sector engaged Cloud Weavers AI to replace this workflow entirely. The solution combined two components of the EnerSmith platform: the web-based Bolt Tensioning System (BTS) for calculation and documentation, and the EnerSmith Mobile App for field data capture via Bluetooth-connected smart hydraulic pumps.

Technicians now work with a tablet running the EnerSmith app. When they connect a smart pump to a joint, the app automatically reads the pump serial number, loads the relevant bolt specification from the BTS database, guides the technician through a multi-pass tightening sequence, and records each pass result with GPS coordinates, timestamp, and technician ID. The completed tightening record is synced to BTS in real time and is available for QA review within seconds.

Within the first six months of deployment, the EnerSmith platform surfaced three anomalies that the previous paper-based system would not have detected until a physical inspection — or until something failed.

The first was a systematic under-tensioning pattern on a specific class of stud bolts from a single batch. EnerSmith's analytics identified that 23 joints across two sites had been tensioned to 87% of the target value. Investigation revealed a calibration drift in one pump unit that had been in service for 14 months without recalibration. The pump was withdrawn, the joints were retensioned, and the supplier was notified.

The second and third anomalies were similar in nature: environmental factors (temperature differential and thread compound contamination respectively) causing systematic deviation in a subset of joints. In both cases, the real-time data visibility allowed the site team to intervene before any joint reached a failure threshold.

Beyond the field execution side of the business, the same client was also struggling with the commercial front-end: responding to RFQs (Request for Quotation) from customers ordering fastener packages for project jobs. Each RFQ arrived as an Excel spreadsheet, a PDF table, or sometimes a scanned image of a hand-drawn BOM.

A commercial team of four was spending 60% of their time manually interpreting these documents, cross-referencing items against in-house manufacturing capabilities, identifying which items needed to be outsourced, and building up a quotation. EnerSmith's WebQuote module, powered by DataSmithAI AI, reduced this process from an average of four hours per RFQ to under eight minutes. The AI reads the RFQ in any format, identifies all fastener line items, checks each against the in-house capability matrix, and produces a structured draft quotation that the commercial team reviews and approves.

The oil and gas industry has been piloting digital transformation for fifteen years. Most pilots stay pilots. The EnerSmith engagement succeeded because it targeted a specific, painful, well-understood workflow — bolting documentation — and replaced it completely, without asking the field teams to run a parallel paper system alongside the new one.

The lesson for any capital-intensive industry considering AI adoption: pick a workflow where the cost of failure is visible and the volume of transactions is high. Start there. Prove the value. Then expand.