Case Study

Senpex Reduces Logistics SLA Risk with Real-Time Data and Generative AI on AWS

At a glance

Senpex partnered with New Math Data to reimagine its dispatch operations around real-time intelligence and Generative AI. The result is a unified logistics intelligence platform built on AWS that enables dispatchers to detect SLA risks instantly, reassign drivers in seconds, and manage deliveries through natural-language interaction.

Industry

Use Case

Workflow and process automation using real-time data streaming and Generative AI

Solution implemented

The value equation

Company Snapshot

Senpex

Senpex is a last-mile logistics provider offering on-demand, same-day, and scheduled delivery with real-time dispatch management.

Location

California, USA

Senpex’s Real-Time Intelligence Platform Transforms Dispatch Decision-Making

Last-mile logistics moves at relentless speed. Customer expectations for on-time delivery leave no margin for error, but real-world variability, traffic, route changes, and driver fatigue create constant risk. Senpex’s dispatchers often lacked a single live view of driver location, route status, and skill/capacity fit, which forced manual, ad-hoc decisions across fragmented systems and raised SLA exposure.

Senpex and New Math Data built a unified, real-time platform on AWS that streams operational data, detects deviations automatically, and enables conversational reassignment, so dispatch can act quickly and confidently.

Problem

Senpex’s dispatchers relied on data scattered across multiple tools to track driver location, route progress, and capacity. When a driver deviated from a planned route or an order required reassignment, operations staff were forced to manually piece together information from different dashboards. These inefficiencies increased SLA risk and placed growing pressure on Senpex’s operations team to deliver more with less.

Senpex needed a production-ready approach to:

Solution

Unifying Operational Data with Streaming into Redshift

Change Data Capture (CDC) from Amazon RDS for MySQL flows through Debezium on Amazon MSK into Amazon Redshift. Materialized views provide near-real-time freshness for dashboards and analysis, without impacting the production database. Provisioned with Terraform, secured by scoped IAM, VPC isolation, and encryption.

Reducing SLA Risk with Event-Driven Deviation Detection

A serverless service on AWS Lambda, DynamoDB, SNS, EventBridge, and CloudWatch continuously evaluates driver telemetry versus planned routes. When deviations or SLA risk are detected, alerts are pushed to Slack for immediate action; CloudWatch dashboards/alarms and DynamoDB history enable visibility and audit.

Accelerating Reassignment with a Conversational Dispatcher Bot

A RAG-based bot (on Amazon ECS behind ALB) uses Amazon Bedrock Knowledge Bases with Redshift to retrieve live data and Claude Sonnet for natural-language reasoning. Dispatchers ask questions like “Who is the closest available driver with capacity to take this order?” and receive context-aware recommendations that factor in geofence, capacity, and route alignment. Sessions are managed in DynamoDB; interfaces include FastAPI endpoints and Slack.

Operating with Observability, Security, and IaC

End-to-end Terraform enables repeatable environments; CloudWatch provides logs/metrics/alarms; encryption in transit/at rest and least-privilege IAM help maintain compliance. The architecture forms a repeatable pattern for future AI logistics services.

Over a focused engagement, New Math Data delivered a production-ready logistics intelligence platform that unified streaming data, automated SLA monitoring, and introduced conversational AI into dispatch workflows.

Related Case Studies

Inspire Clean Energy

Streamlined Data Processing and MLOps with Spark

Vertically Integrated Utility Company

Migration of data systems and applications to AWS.

Ready to Transform Your Logistics Business?

See how New Math Data can transform your logistics business with AWS-powered innovation.