Back to Our Work
Logistics & Supply ChainSeries A

Velthari

How we helped Velthari transform their technology.

The challenge

Velthari was building an AI-powered freight matching platform in Amsterdam to connect European shippers with carriers across 12 countries. Their 5-person team had validated demand with a basic web portal, but they had no mobile presence — and 70% of their carrier base operated from phones on the road. Carriers were calling in to accept loads because the web app wasn't usable on mobile, costing Velthari 2 support reps and killing match speed. On top of that, their matching algorithm was a simple rules engine that couldn't account for route optimization, carrier reliability history, or real-time pricing fluctuations. With EUR 9M in Series A funding and contracts pending with three major Dutch logistics firms, they needed a native mobile experience for both platforms and an AI matching engine that could outperform manual dispatch — all within 5 months.

Our solution

We embedded four engineers with the Velthari team for 20 weeks — two full-stack engineers focused on the mobile apps, one AI/ML engineer, and one senior backend engineer. We built native iOS and Android apps using React Native with shared business logic, giving carriers real-time load notifications, one-tap acceptance, GPS tracking, and in-app document upload for proof of delivery. In parallel, we redesigned their web platform in Next.js with a shipper dashboard for real-time shipment visibility and analytics. For the AI matching engine, we trained a gradient-boosted model on 14 months of historical match data — factoring in carrier reliability scores, route efficiency, pricing patterns, and delivery time predictions — then deployed it on AWS SageMaker with a real-time inference endpoint. We also built a feedback loop where completed shipment outcomes continuously improved model accuracy. The entire stack was deployed on AWS with ECS, CloudFront, and a PostgreSQL database on RDS Multi-AZ to handle cross-border latency requirements.

The impact

8x faster
Match Speed
AI matching engine reduced average time-to-match from 4.2 hours (manual) to under 30 minutes
92%
Carrier Adoption
92% of active carriers moved to mobile apps within 6 weeks — support calls dropped by 74%
94%
Match Accuracy
AI model achieved 94% successful match rate vs. 61% with the previous rules engine
+180%
Revenue Growth
Platform GMV grew 180% in the first quarter post-launch, unlocking all 3 pending enterprise contracts

Our carriers were calling in to accept loads — we were running a tech company on phone calls. Kavora built our mobile apps, web platform, and an AI matching engine that outperforms manual dispatch by a factor of 8. We went from duct tape to the most advanced freight platform in the Benelux.

Lars de Vries
CTO, Velthari