Lighthouz AI

The interesting thing about Lighthouz is what it refuses to do. It does not replace your TMS — it reads from it, and its audit logic depends on clean load data being there. That constraint is also its clarity: brokers already have McLeod, Turvo, MercuryGate, Tai, 3PL Systems, Brokerware, Revenova, Sunnybrook, or similar, and Lighthouz plugs in alongside rather than asking for a migration. The vendor explicitly says no rip-and-replace. The agent set splits cleanly across AP and AR. On AP you get four agents — Freight Bill Processing & Audit, Rate Verification, Payment Status Response, and Dispute. The audit agent runs 45+ checks per document across linehaul, fuel surcharge, detention, lumper fees, and accessorials, and the live activity feed shows what it did line by line ("linehaul match ✓, FSC match ✓, detention flagged ✗"). On AR you get the Shipper Invoicing Agent, which pushes invoices to customer portals within hours of delivery, and two collections agents — one that monitors portals like Cass and Uber Freight, and one that runs dunning sequences trained on your own SOP. The AI-ranked work queue and AR risk dashboard turn a date-sorted aging report into a priority list. Strengths: mode-native FTL and LTL handling (reweighs, reclasses, and accessorial disputes are where generic AP tools fall over), a finance-grade compliance posture with SOC 2 Type II, encryption in transit and at rest, role-based access, audit logging, and tenant-scoped isolation. Weaknesses to weigh: the headline numbers (70–85% no-touch, 12-day DSO reduction, up to 80% lower back-office cost) are vendor-reported and not independently audited, so they are claims to test, not results to bank. The vendor does not name the underlying AI models. The product targets FTL and LTL only — parcel and intermodal operations are out of scope. Small brokerages under roughly 100 bills a month likely won't generate enough volume to justify onboarding. And because the audit engine depends on TMS data quality, a broker with messy load records will see the no-touch rate fall well below the advertised range. Where it fits: mid-market and growing brokerages where AP/AR headcount would otherwise have to grow with revenue, and finance teams that need SOC 2 controls before they'll route money through an AI system. Where it doesn't: parcel-only shops, intermodal-only shops, teams that need on-premise deployment, and anyone without a TMS to audit against.