One loop, run continuously
AVIR reads your systems, simulates the options, ranks them, commits the best action, and measures what actually happened, then calibrates. Continuously.
Six stages, one loop
The same loop runs on every decision, whatever the operation.
Read
AVIR connects to the systems you already run and reads them continuously (orders, inventory, messages, schedules) into one live model of your operation: what exists, what depends on what, and what constrains it.
Simulate
It projects thousands of possible actions forward and models what each would do. Not a single forecast: the full range of options.
Rank
Every option is scored against your goals (cost, risk, time) and ranked, so the strongest choices rise to the top.
Commit
AVIR writes the chosen action back into your systems: the action itself, not a recommendation to execute by hand.
Observe
It records what actually happened next to what it predicted: an outcome ledger. Every decision is measured, not assumed.
Calibrate
The gap between predicted and actual feeds back in, so the next round is sharper. The loop improves the longer it runs.
Plenty of systems will recommend an action, and some will execute it. Almost none record what they predicted next to what actually happened. That ledger is what makes the next recommendation better than the last one.
Manual and Auto
You decide how far the loop runs on its own, set per action type.
Manual
AVIR ranks the options and waits. Nothing is written back to your systems until a person approves. You sign off on every commit.
Auto
AVIR commits the chosen action within limits you set (value thresholds, action types, guardrails). You review the outcome ledger after, not before.
Routine, low-risk moves run on Auto; anything material is held for sign-off.
Fork any scenario
The loop runs continuously. You can also drive it yourself. Fork any scenario, set your own levers, and run it. It uses the same simulation engine and the same cost model as the automatic loop, so a manual answer and an automatic one never contradict each other.
Live with aerospace operators today
Real-time repair lifecycle intelligence across a major carrier’s cabin fleet: from intake to dispatch, every asset tracked, every TAT measured.
Deterministic asset-state control for a major carrier's cabin operations: per-asset state tracking, closed-loop compliance, station-level telemetry, live in production.
Demand-change cascades that took a week now resolve in under a minute: BOM coverage, supplier capacity, hub inventory, and exposure in one pass.
API-first, no legacy connectors
Data ingestion
AVIR connects to your systems via standard APIs: REST and webhooks. No custom ABAP, no file-based integrations, no middleware to maintain.
Intelligence processing
Every signal is read against the live model of your own operation (your assets, suppliers, schedules, and constraints) by purpose-built models, not a general chatbot.
Enterprise-grade data handling
Data isolation
Every customer's data is logically isolated. No cross-tenant data access. Regulated workloads can run on dedicated infrastructure.
Encryption
AES-256 at rest, TLS 1.3 in transit. API keys and credentials stored in encrypted vaults, never in application code.
Access control
Role-based access with programme-level granularity. Restricted-programme data limited to cleared users. Full audit log of every access event.
Compliance posture
SOC 2 Type II in progress. GDPR compliant. Data residency controls available. DR and backup evidence packs available on request.
What we do and don’t do with your data
We do
We don't
Get early access to new capabilities as they ship.
Need more detail?
Technical docs, architecture diagrams, and security questionnaire responses on request.