AI Is Eating Logistics
Ryan Petersen, founder and CEO of Flexport, explains how AI and Machine Learning are being implemented to revolutionize the multi-trillion-dollar logistics industry. He details specific applications, from ML models that optimize container routing to LLM agents that automate communication, and discusses the cultural and strategic shifts required for a large company to embrace AI-driven, bottom-up innovation.
Flexport’s business model is described as “scale economies shared,” aiming to leverage technology and scale to drive down costs for customers. The core thesis is that as the company grows and integrates more automation, it can reduce its costs and pass those savings on to its clients, creating a virtuous cycle of growth. The tangible goal is to make ocean container shipping 8-10% cheaper over the next few years, with AI being a significant driver of this efficiency.
The Journey of AI Adoption and Innovation
The adoption of AI, particularly since the launch of ChatGPT in late 2022, has transformed the company’s innovation culture. This shift is most visible in Flexport’s internal hackathons.
- From Top-Down to Bottom-Up: While traditionally a founder-driven, top-down approach was necessary to align the company, the rise of accessible AI tools has unlocked a wave of bottom-up innovation. It’s now recognized that engineers and operators on the front lines are often best positioned to identify and build the most impactful AI applications.
- Hackathons as a Product Pipeline: Internal hackathons, held twice a year, have evolved from exercises into a primary source for the product roadmap. Previously, only a few projects were AI-related; now, an estimated 90% are based on LLMs. These projects are increasingly being funded and developed into real features, prompting a strategic shift to schedule hackathons just before the main roadmap and budgeting cycles.
Key AI Applications and Their Impact
Flexport is deploying AI across its operations, impacting both internal efficiency and the customer experience.
Operational Optimization and Automation
- Container Planning: A long-standing machine learning model optimizes container routing. By analyzing contracts, sailing schedules, and variability, the model determines the most efficient path for shipments. This system has already saved 2% of ocean freight spend while simultaneously improving transit times by 20%—a significant achievement in an industry where speed and cost are usually a trade-off. A key function is its ability to re-optimize constantly, for instance, by pulling a future shipment forward to fill a slot from a last-minute cancellation, a task impossible to perform at scale with human operators.
- Automating Routine Work: LLM agents are being deployed to handle tasks that were previously manual or too costly to perform consistently.
- An LLM agent uses both email and voice calls to verify warehouse addresses and confirm delivery appointments, reducing the costly errors that arise from bad address data.
- The company is on track to increase the automation of manual work from 20% at the start of the year to 50% by the end, with a future goal of 90-95% automation.
Enhancing the Customer Experience
- Natural Language Data Queries: A hackathon project evolved into a feature that allows customers to query their supply chain data using natural language. This eliminates the need for SQL or pre-built dashboards, directly answers customer questions with graphs and charts, and has freed up approximately 25% of account managers’ time previously spent on generating reports.
- Customer Sentiment Analysis: An AI model analyzes messages within the Flexport platform to detect customer dissatisfaction. If the model identifies negative sentiment or emotion in the communication, it automatically triggers an escalation to a manager, allowing for proactive problem resolution.
Empowering the Workforce with AI Skills
Flexport has instituted a program to train non-engineers in AI and coding. Participants dedicate one day a week for 90 days to an “AI bootcamp” to learn how to build their own tools and automations. The goal is to empower domain experts—the people who best understand the repetitive aspects of their jobs—to automate their own workflows, turning them into low-code developers.
Economic and Philosophical Outlook
The automation driven by AI has profound economic implications. Labor accounts for roughly 10% of the cost in the freight forwarding industry. By automating this layer, Flexport believes it can reduce the overall cost of shipping by 8-9%. This could have significant ripple effects, potentially boosting global trade and contributing to GDP growth.
This technological shift is compared to historical moments like the “Axial Age,” when the invention of coins depersonalized transactions and required new philosophical frameworks. Similarly, AI will change how society operates, but it won’t eliminate human ambition. The fundamental human desire for “more” will ensure that new jobs and new wants emerge as old ones are automated away.
The Future Vision for Flexport
The long-term vision is for Flexport to become a global logistics utility. The goal is to create a platform where anyone can ship anything, anywhere, through any mode of transport, all orchestrated via code and APIs. Logistics should be as reliable and invisible as the electrical grid, allowing businesses to focus on their core purpose: making great products and talking to their users. To achieve this, Flexport plans to have its own physical presence in countries covering 95% of all container trade by 2028, ensuring it can control and automate the end-to-end process.