Applicability
Identify the exact aircraft and relevant configuration, equipment, mission, role, and operating context.
Safety and trust
Trust starts before the first tap.
A reliable runner and reliable content are inseparable. Cockpit Checklist is designed to keep procedure state, applicability, provenance, and the limits of automation visible.
CC
ASSURANCE PRINCIPLE 01Never claim more certainty than the system has.Core assurance principles
Make the important truth visible.
Provenance
Carry source, revision, section, claimant, and asserted authority with the released content.
Completion truth
Keep required, deferred, and exceptional states visible. Unresolved does not become complete.
Recovery
Preserve local session state through interruption, app restart, device reboot, and connectivity loss.
Change control
Use immutable releases, semantic differences, acknowledgements, staged rollout, recall, and rollback.
Honest verification
A tap or voice response records a pilot declaration; it does not sense the physical aircraft.
Human factors
Test glare, darkness, vibration, noise, gloves, one-hand use, aging vision, and workload.
Privacy
Keep personal history private by default, with explicit sharing and purpose-limited organization access.
Plain-language trust labels
The label names who stands behind what.
“Verified” is too vague for high-consequence content. Trust labels identify the claimant and evidence while preserving the pilot’s ability to inspect source and effectivity.
Created or adapted by an individual; no outside review implied.
Linked to an identified source, section, and revision.
Released through a named organization workflow.
Origin and integrity asserted by the identified publisher.
Content governance
AI can help prepare a draft. It cannot become the authority.
Scanning, extraction, classification, and comparison can reduce manual work. Every result remains a draft until reviewed through the required human process against the applicable source.
- OCR preserves page and source anchors where possible
- Extracted steps remain visibly unreviewed
- Semantic checks flag ambiguity and inconsistency
- Qualified humans determine meaning and release status
- No model output silently changes a live procedure
Validation before claims
Test the failure, not just the happy path.
Coverage and intended-use claims advance only after the relevant content, cockpit, device, and organization workflow have passed their own gates.
Representative pilots
Students, CFIs, owners, renters, professional single pilots, multi-crew pilots, and rotorcraft operators.
Representative workload
Interruption, glare, darkness, vibration, noise, gloves, mounting position, time pressure, and older eyes.
Representative breakdowns
Offline startup, stale release, wrong aircraft, app crash, reboot, low storage, bad recognition, and multi-device conflict.
Mechanism-level evidence
Omission, incorrect confirmation, resume accuracy, heads-down time, workload, and fallback success—not engagement.
IMPORTANT BOUNDARY
A support system—not a blanket approval.
Cockpit Checklist does not replace the approved AFM, POH, RFM, operator procedures, pilot judgment, or deployment-specific authorization. Content and intended-use claims apply only to the aircraft, configuration, operation, and release explicitly identified. No blanket FAA, OEM, or operator approval is implied.
Human in the loop
The product never pretends to see the aircraft.
Unless a separately approved aircraft integration provides a defined signal, the system knows only what the pilot declares. Voice recognition creates a pending attestation, not proof of a switch, value, or configuration.
PILOT OBSERVESPhysical aircraft state
→SYSTEM RECORDSPilot’s declared response
≠ automatic aircraft sensingHelp us validate the hard parts
Bring us the interruption, edge case, or release problem that worries you.
We’re building the research and design-partner network across fixed-wing, rotorcraft, training, and professional operations.