From Manual Outreach to AI-Powered Matching: Commerce Architects Builds Conversational AI Platform for Freight Logistics

Commerce Architects
19 hours ago
2 min read

When Freight Matching Meets AI: Building a Conversational Calling Platform

A freight logistics startup needed to build an AI-powered outbound calling solution for matching independent shipping operators with available freight loads. The company had a short timeline to get an MVP solution in place to start connecting with existing contacts and demonstrate viability to potential investors. With an aggressive timeline to secure customers and investment, the company engaged Commerce Architects—an AWS Advanced Tier Services Partner—to develop its Gen AI MVP.

The Challenge: Automating Freight Matching at Scale

The startup team recognized the opportunity to streamline how independent drivers are matched to available freight jobs. They wanted to build an outbound calling platform that could autonomously engage drivers, answer their questions about loads, match them with jobs, and scale rapidly. For example, when a customer needed to pick up freight from a distribution center and move it to a specific location, the load could be registered with the platform. The system would automatically call a pool of drivers to evaluate their interest in transporting the load.

The Solution: Conversational AI-Powered Outbound Calling

For the platform, the Commerce Architects team leveraged Amazon Connect voice-to-text translation for driver phone interactions, allowing an LLM backend to infer answers to questions or comments from the caller about the load in question. After Amazon Connect and Amazon Lex convert voice to text, Amazon Bedrock processes the text via an LLM foundational model that interprets the questions to infer what the caller is asking and sources answers about the load from backend APIs. This supports a conversational interface where the prospective driver can learn about the freight in question and then request to be transferred to an operator to book the job.

Architecture Diagram:

Results: Rapid MVP Delivery and Accelerated Time to Market

We delivered the MVP in six weeks through tight collaboration and regular testing cycles. Each demo review kept us aligned with the startup's objectives, while continuous testing gauged progress and informed rapid adjustments.

What we built:

Complete Amazon Connect environment with intelligent call routing
Natural language conversation flows using Amazon Lex
Serverless Lambda functions connecting to Amazon Bedrock for AI-powered responses
Backend integrations pulling live freight data