Vision AI reads every inspection photo and flags what your team would have missed.
Every vessel. Every handoff. Every photo. A nightly batch that compares today to yesterday and surfaces wear, damage, and drift.
The problem
What operators actually say.
“Damage disputes turn into he-said / she-said weeks after a trip.”
“Inspection photos pile up on a shared drive and nobody actually reads them.”
“Wear-and-tear accumulates invisibly until a part fails at the worst possible moment.”
How it works
Three steps from install to live.
Step 1
Capture on a phone
Dock staff take photos in the same workflow they already use.
Step 2
Compare to last inspection
The model compares current photos to prior baselines for each vessel.
Step 3
Review the flags
Severity-scored findings land in the morning report, ready for the service board.
What's inside
Everything in Fleet Inspection AI.
Per-vessel baselines
Each boat has its own photo history, not a generic damage model.
Severity scoring
Three-tier severity with confidence scores; you decide what threshold escalates.
Nightly batch
Runs after the last dock closes; morning report lands before staff arrive.
Email + dashboard
Flags appear in Hub Portal and in a plaintext summary email.
Downstream to AMS
High-severity findings auto-open tickets in MarineOps AMS.
The AI moment
The AI moment
Claude Vision compares previous and current inspection photos per vessel; damage and wear severity scoring; nightly batch runs; email reports. Photos and annotations are never used to train foundation models.
Built on real operations
Proven at Freedom Boat Club Northeast Florida.
At FBC Northeast Florida, time-to-flag dropped from weeks to the next morning; damage disputes with members are down materially.
SOC 2 Type I in progress. AES-256 at rest, TLS 1.3 in transit. Customer data is never used to train foundation models.