// flagship
Narya Command™
automation application
The 100× engineer maximizer for your ops team. Workflows on a schedule or trigger. Deterministic connector steps for known work, AI only when judgment is required, plus WAIS site memory and preview/confirm/rollback when writes touch SCADA.
// flow
Known work runs as deterministic connector steps → ambiguous steps get an AI prompt with strict schema → preview before writes → you confirm → WAIS captures the lesson → full run log retained
// the signature workflow
The five-minute loop your team will not run manually again
Narya Command is the automation app your team runs operational work in. Workflows trigger on a schedule, an inbox event, or an operator. Steps query CygNet, write to Azure DevOps, classify email with AI, and notify the engineer when triage is ready.
- 01Schedule — Run this workflow every five minutes
- 02Email scan — Check inbox for new CygNet help requests
- 03AI analyze — Classify the request and extract device, symptom, urgency into strict JSON
- 04Azure DevOps — Open or update a work item with the request and assigned engineer
- 05CygNet diagnostics — Query points, recent alarms, and trend context for the device
- 06Write findings — Update the Azure DevOps work item with the diagnostic summary
- 07Notify the Engineer in the Command Center — A new item needs to be triaged
The run workbench your team operates from
Design workflows on a canvas, configure each step, and inspect run logs — cross-system triage from email to Azure DevOps to SCADA investigation with human gates where it matters.
command center
Notifications, waiting runs, and recent outcomes in one view.
workflow canvas
Step sequences with connector actions, validation, and AI prompt nodes.
run log
Per-run status, trigger, duration, pause points, and outcomes.
Three connectors. One run engine.
CygNet, Email, and Azure DevOps built in today. One run engine, one safety model, one audit trail — whether the workflow touches SCADA, mail, or engineering systems. See all connectors →
CygNet
Deep SCADA coverage — preview/apply writes, device clone, bulk updates, alarm workflows, and the ODBC query designer.
Outlook Classic or Microsoft 365 — read messages, extract attachments, classify intent, and route operational mail into workflows.
Azure DevOps
Create and update work items, add comments, and tie engineering triage back to the same run audit trail as your SCADA steps.
future
Snowflake
Analytics and warehouse queries inside governed operational workflows.
The intelligence layer that gets smarter at your site
WAIS — the Westmarch Adaptive Intelligence System — is built into Narya Command. Multi-path retrieval (structured ontology, lexical recall, semantic recall, all reranked for relevance). Memory layers for lessons, conventions, and aliases. Authority-weighted reinforcement so senior admins promote knowledge faster. Golden Rules — immutable product-level truths — never decay. See how WAIS works →
// two surfaces you'll use every day
Chat and query designer
inside the automation app.
Workflows handle the recurring cross-system loop. Chat and the query designer handle ad-hoc CygNet administration and data investigation when you need a human in the loop.
chat operations
query designer
token visibility
// the cool stuff underneath
Built for production CygNet
Real operational throughput needs more than a chat box. Rollback with conflict detection, real-time token visibility, spreadsheet-driven bulk ops, full device clones, reusable skills, and a learning layer that compounds with every correction.
// example prompts
Conversational administration
The point is not chatting for fun. The point is getting real CygNet work done faster and with less risk — and capturing what your team learns along the way.
- "Show me all TotalFlow devices with Oil Meter facilities"
- "Clone TESTDEVICE to create a new compressor site called AUSTIN125H"
- "Change facility attribute 13 to 'OIL' for all PLUNGER facilities on Route 19"
- "Deactivate every facility listed in this spreadsheet and show me the preview first"
Describe the data. Get the SQL. Save it for next week.
CygNet ODBC is powerful and unforgiving — table names buried in documentation, flag columns with non-obvious values, and joins that change shape per site. The Query Designer is built around all of it.
Schema-aware from the first prompt
Type-ahead schema search, automatic column discovery, and live sampling of flag columns (real distinct values like Y / N instead of guessed booleans) get injected into the prompt so generated SQL matches your actual data.
Conversational refinement
Ask in plain English; Narya generates SQL, asks clarifying questions when intent is ambiguous, and iterates inline. Tab history shows the latest result query, not the first exploratory one.
Saveable, parameterized queries
Save with {{variable}} placeholders. Re-run with new inputs next week. Every team has the same five questions they ask Monday morning — make them one-clicks.
Result tools that match the workday
Column sort and filter, pinned tabs for parallel investigations, execution history, configurable row limits, SQL formatting, and export to CSV, Markdown, or JSON. ODBC writes get a separate Run Write button so reads and writes never get confused.
Preview → confirm → rollback (with optional VHS history)
Every write shows a before/after preview. You approve or deny. Changesets capture an independent audit trail and support undo for 30 days. Rollback re-reads current values, flags conflicts where another operator has changed something since your write, and only reverts what's safe — no silent overwrites of someone else's work.
For operations that touched the historian, opt in to VHS history rollback and the changeset will restore historian values alongside the configuration revert — so a bad bulk update doesn't leave stranded data in VHS. CygNet ACS stays the authority for permissions and audit; we don't replace it.
Cost transparency built into the chat
AI conversations are not free, and pretending they are is how budgets get blown. Narya tracks input tokens (cached and uncached), output tokens, model used, and estimated cost on every turn — persisted per user, per conversation, and per site so finance and engineering can both see what's actually being spent.
- > Live context-size indicator with amber → orange → red warnings as a conversation grows past 50K, 100K, and 150K tokens.
- > One-click compaction summarizes long conversations so quality and cost stay under control without losing the audit trail.
- > Aggregate dashboards for tokens-per-user, tokens-per-site, and cost per billing period.
- > Cache hit rate visibility so you can see prompt caching working — and catch it when it stops.
// vs generic ai
Narya Command vs DIY connectors
Generic AI plus an API can answer questions. Narya Command is built to carry operational work all the way through review, execution, and into your team's growing chronicle.
| Narya Command™ | Generic AI + API | |
|---|---|---|
| CygNet domain knowledge | Months of battle-tested patterns, plus WAIS learning per site | You build it from scratch |
| Adaptive learning per site | WAIS — corrections become lessons, lessons become Skills | None — every conversation starts cold |
| Write safety | Preview / confirm / rollback on writes | Roll your own |
| Reusable playbooks | CygNet skills triggered by slash commands, built on the open Agent Skills format | Typically unavailable |
| Cross-admin knowledge sharing | Centralized WAIS — every admin's corrections benefit every other admin | None |
| Deployment | Local Windows app + optional centralized WAIS server | Varies |
// get started
Try the beta
Sign in, get an NRYA-format beta license, and download the Windows installer. Put Narya Command™ in front of your team now for scheduled workflows, cross-system triage, Skills, and CygNet administration while pricing for general availability is finalized.