Skip to main content

// about

About Narya

We build practical AI systems for oil and gas teams that need less administrative drag, better operational throughput, and a stronger foundation for production decision support.

Our Mission
To build practical AI systems for oil and gas operations, giving operators, technicians, and engineers tools that reduce SCADA friction while improving how production decisions get made.
The Problem
Traditional SCADA systems force teams through repetitive navigation, brittle manual edits, and too much screen-by-screen work just to get routine operations done. That slows down the field, raises risk, and steals time from higher-value decisions.
Our Solution
Narya builds Narya Command (the governed automation harness), WAIS (the intelligence layer that learns your site), and connector plugins with CygNet as the first expert wedge. Together they automate cross-system operational work with preview/apply safety and a growing site chronicle.

// the names

Why Narya, why Westmarch

Narya — the Ring of Fire

Narya was worn by Gandalf. Its power was to inspire courage and action in others — not to control or dominate. That maps directly to the product thesis: the AI assists and enables the CygNet administrator. The human decides and acts. We chose the name to avoid the “Copilot” baggage and to position Narya as a product family with CygNet as its first target domain.

Westmarch — the growing chronicle

The Red Book of Westmarch is the chronicle that compiled the hard-won knowledge of many journeys, written and added to over generations. WAIS is your CygNet's growing chronicle. Every correction, every convention, every learned pattern is compiled into the collective knowledge of your site — sharper every week, shared across every administrator when you deploy it centrally.

// north star

80% resolvable from accumulated knowledge

The long-term goal is a coverage-driven triage layer: 80% of incoming operational problems at a mature site resolvable from your accumulated CygNet knowledge alone, without fresh learning every time. The 80% figure is aspirational — it expresses the ambition that the system should eventually carry most of the routine operational load. The remaining ~20% is where novel situations, genuine engineering judgment, and continued learning live. We instrument coverage with a measurable coverage_ratio — distinct (device_type × action × outcome) tuples observed at a site, divided by the known-possible surface for that site. A real metric, not a vibe.

Higher coverage never means looser safety. Preview, confirm, changeset, rollback, and human-in-the-loop approval remain the default at every autonomy level — including in the future on-prem Triage Agent. See where this is going →

// vision

The Future of SCADA

We believe AI creates its first real value in SCADA by removing administrative drag. Faster device setup, easier bulk changes, better search, and more efficient daily workflows mean teams can spend less time clicking through screens and more time solving operational problems.

We also see major opportunity in well production strategy — including plunger optimization and production decision support. Our approach to machine learning is to build custom models trained specifically on well production data so recommendations are grounded in the realities of field behavior, not just generic AI patterns.