AI on every screen —
wired into the operation, not bolted on.
A My Day brief for every seat, a dispatch watch that flags FAR-117 and weather risk, plain-English explanations of maintenance and fatigue risk — each one wired into a real screen, written from your own data, not a demo. Advisory and opt-in: set the key and it lights up; leave it unset and it stays silent.
Launching the first quarter of 2027.
7 sim sessions and 2 check rides today. Capt. Reyes is blocked — a medical expired yesterday, so hold his check ride until it's cleared. N4821G is AOG on an open airworthiness directive, and 3 credentials lapse this week. The rest of the board is clean.
My Day card on the dashboard.
What it is
Useful where the work happens
- A “My Day” brief for every seat — director, dispatcher, pilot, flight attendant — written from that person’s own schedule, work, and credentials.
- A dispatch watch that reads today’s legs and flags FAR-117 crew-legality risk, grounded aircraft, open MEL deferrals, and weather.
- Plain-English explanations of maintenance and fatigue risk — the score stays deterministic, the AI only narrates what’s driving it.
- A pre-session instructor brief that summarizes a trainee’s recent progress and expiring endorsements before a check.
- Advisory pairing-optimization suggestions inside Part 121 bidding — a starting point a crew planner reviews, never an auto-decision.
- Opt-in by design: set an API key and it lights up; leave it unset and every AI surface simply hides itself.
How an operator actually uses this
It's 6:50am. The chief pilot opens AviationAlley before the first crew brief and reads three sentences: which sessions and check rides run today, who's blocked, and what's grounded.
A captain's medical expired yesterday, so his check ride is on hold. A tail is AOG on an open AD. Three credentials lapse this week. Each is a chip — one tap opens the exact filtered screen.
By 7am the day is triaged: the medical is flagged to the scheduler, the AD work order is reassigned, and the credential reminders are out — none of it dug out of five tabs.
Plain and honest — the brief reflects what's in the data, nothing more.
Jump to a feature
Every AI surface, broken down
Jump to any surface — each one is detailed below with a look at the actual screen.
Deep dive · 01 — Grounded in your data
Built from your real ops data, not LLM guesswork
The My Day brief and the dispatch watch don't ask a model to recall facts about your operation. They read the database first — today's sessions, overdue compliance, critical work orders, your credentials — assemble those structured numbers, and only then ask the model to turn them into a few plain sentences. The figures the brief was built from are shown right alongside it, so every claim traces back to a real row.
- Structured JSON of today's real numbers is built before the model is called.
- The model rewrites those numbers as plain English — it isn't asked to invent facts.
- The underlying figures are shown next to the brief, so claims are auditable.
- Refresh regenerates from the latest data whenever you want it.
7 sim sessions and 2 check rides today. Capt. Reyes is blocked — a medical expired yesterday, so hold his check ride until it's cleared. N4821G is AOG on an open airworthiness directive, and 3 credentials lapse this week. The rest of the board is clean.
Deep dive · 02 — From brief to work
Action chips that deep-link to the work
A brief that just describes the day isn't enough. Each item the brief raises becomes an action chip, and tapping it opens the precise filtered view — the expiring credentials, the overdue compliance, the open work order. No re-searching, no guessing which screen; the chip takes you straight to the rows that need attention.
- Every flagged item is a chip, not just a sentence.
- A chip opens the exact filtered screen — e.g. credentials filtered to expiring this week.
- Triage the morning without digging through five tabs.
- Works the same for each seat: director, dispatcher, pilot, flight attendant.
Deep dive · 03 — Explainable, not a black box
Risk you can trace, not a black box
Predictive-maintenance AOG risk and crew fatigue risk are scored deterministically — from airworthiness-directive posture, the inspection clock, utilization, age, and (for fatigue) FAR 117 duty load blended with self-reported signals. The AI never produces the number. Its only job is to explain, in plain English, which of those already-computed factors are driving the score and what to watch. This is explainable math with a narration layer — not an AI prediction.
- The risk score is computed deterministically — same inputs always give the same number.
- The AI narrates the contributing factors; it does not introduce new risk signals.
- Each factor carries its own point contribution, shown next to the explanation.
- Explicitly not a prediction — the math is the source of truth, the words are the gloss.
Deep dive · 04 — What needs attention now
A dispatch watch that flags what needs attention
The ops watch is the dispatch desk's version of My Day. It reads today's legs and their status, any grounded aircraft or simulators, open MEL deferrals with their expiry, crew-legality (FAR 117) risk, and weather flags resolved for the airports actually on today's plans — then synthesizes one short read of what needs attention right now. Like every surface here, it's advisory and built only from data already in the system.
- Reads today's legs, statuses, and grounded equipment.
- Flags FAR-117 crew-legality risk and open MEL deferrals nearing expiry.
- Surfaces weather flags for the airports on today's plans.
- Advisory only — the dispatcher and PIC remain the decision-makers.
Deep dive · 05 — Opt-in by design
Degrades gracefully — set the key, it lights up; unset, it's silent
The AI surfaces are gated on an Anthropic API key. Set it, and the briefs, the ops watch, and the risk explanations appear. Leave it unset — local dev, a demo center, or an operator that opts out — and every AI card simply hides itself while the rest of the platform runs exactly the same. If a call ever fails, the surface hides rather than throwing. AI never blocks the work.
- One environment key turns the whole AI layer on or off.
- Unset key → every AI card hides; the platform is fully usable without it.
- A failed AI call hides the card instead of erroring — never a 500.
- Opt-in, not on-by-default: you choose whether AI runs at all.
See the AI on your own operation.
Advisory, opt-in, and grounded in your data. Request a demo or join the waitlist.