How is Winda different from traditional integrity software?
Legacy tools show you data. You still decide what it means. Winda closes the loop from field signal to action. Ranked, explained recommendations, not another dashboard. The stack is modular. Software only, software plus hardware, or the full stack with our engineers embedded. Every recommendation carries evidence, standards anchor and confidence. An engineer approves before anything moves.
What is Winda Blueprint?
Blueprint is our design-phase product. We turn a project's early design data, the FEED package, process flow diagrams and standards, into finished engineering deliverables like material selection reports and datasheets. The design lives as one model, so changing a source value regenerates every document that depends on it. Standards are held as hard limits, and an engineer approves every step. Blueprint is early, and we are building it with a first EPC partner now.
What is Winda Atlas, and why does it matter?
Atlas is the data layer of our stack. A semantic model of oil and gas integrity, built on your data. Every asset, damage mechanism, inspection and action encoded, with corrosion physics built in. When the AI reasons about your pipeline, it is calling domain models, not guessing from text. Atlas learns from every verified outcome. Accuracy compounds cycle after cycle, on your data, owned by you.
Does it work with our existing SCADA / ERP / CMMS?
Yes. The stack ingests inspection, SCADA, sensor, lab and document data from any source. Maximo, SAP PM, OSIsoft PI, Meridium and Power BI plug in and read. Open architecture, no rip-and-replace.
Do you support on-premise deployment? How is data secured?
Yes. On-premise and in-country hosting are supported, with export behind an explicit approval workflow. We deploy every product fully air-gapped, and your data trains only your models. From any CMMS, historian or document pile to a working decision queue. OT integration is read-only by design. No write-back to the control layer, air-gap supported. TLS 1.3 in transit, AES-256 at rest, customer-managed keys, zero-trust IAM. Winda holds IEC 62443, ISO 27001 and SOC 2 Type II. Tenant isolation is contractual. Your data trains your models, not a shared pool.
Can the AI give wrong answers?
Any model can be wrong, so we engineered for it rather than around it. Recommendations pass confidence-threshold gates. Every number carries a reasoning trail, source document, standard clause, calculation lineage. The loop has a mandatory human approval stage where your engineers make the call with the evidence in front of them. Below the confidence threshold, the system declines to answer.
