AI Is Eating Logistics

Key Concepts

Logistics as a Scale-Driven Industry: The core principle that larger scale leads to lower costs in logistics.
AI in Logistics: The application of artificial intelligence to optimize various aspects of the logistics process, from customer experience to operational efficiency.
Economies of Scale Shared: Flexport's business model of leveraging scale to reduce costs and passing those savings onto customers.
LLMs (Large Language Models): A type of AI that excels at understanding and generating human-like text, used for tasks like parsing unstructured data and customer communication.
Hackathons: Internal company events used to foster innovation and identify potential AI-driven product features.
Bottom-Up Innovation: Innovation driven by individuals or teams within the company, often stemming from practical problem-solving.
Top-Down Directive: A leadership approach where management sets clear goals and direction for the company.
Freight Forwarding: The service of arranging shipments for individuals or corporations to get goods from the manufacturer to a market, customer, or final point of distribution.
Human-in-the-Loop: The concept of requiring human oversight or intervention in AI-driven processes, often for regulatory or liability reasons.
Axial Age: A historical period characterized by significant philosophical and religious developments, paralleled with the emergence of impersonal transaction systems like coinage.

Flexport's AI-Driven Transformation in Logistics

Introduction to Flexport and AI Implementation

Flexport is a global logistics company built on a modern tech stack, aiming to facilitate the movement of cargo across various transportation modes (air, truck, rail) efficiently and cost-effectively. Ryan Peterson, founder of Flexport, highlights the significant role of AI in achieving these goals. The company's AI initiatives span from enhancing customer user experience and optimizing container loading to automating manual tasks previously handled through email or phone.

AI's Impact on Cost and Efficiency

A key takeaway is Flexport's projection to reduce ocean container shipping costs by 8-10% over the next few years, with AI being a major contributor. Specifically, their AI for planning has already saved 2% of their ocean freight spend while simultaneously improving transit times by 20%. This achievement defies the typical trade-off between speed and cost in logistics.

The "Economies of Scale Shared" Model

Peterson likens Flexport's business model to Costco's, emphasizing that as the company grows larger, it becomes cheaper to operate. This increased efficiency, driven by automation, is then shared with customers, encouraging further volume and reinforcing the scale benefits. Flexport, with $2 billion in annual revenue, is described as "just getting started."

The Evolution of AI Adoption Post-ChatGPT

The launch of ChatGPT in November 2022 marked a significant inflection point for Flexport. Peterson admits to being personally "obsessed" with LLMs since then and has actively pushed for their adoption within the company, fostering a sense of urgency to avoid becoming a "boomer company." He notes that Flexport, as one of the few large logistics companies founded since the web browser, must lead in AI adoption.

Advantages of Incumbents in AI Adoption

Peterson outlines three key advantages that established companies like Flexport possess in leveraging AI:

Scale of Data: Access to vast amounts of historical and operational data for training AI models.
Domain Experience: Deep understanding of industry problems and the ability to identify valuable AI applications that might be too niche for startups.
Distribution: The ability to deploy successful AI products across a large existing customer base immediately, unlike startups that need to build trust and acquire data.

Flexport's Tech Stack Advantage

Despite being a young company relative to the industry, Flexport's self-built tech stack allows for seamless integration of AI. In contrast, many competitors rely on third-party, off-the-shelf solutions, making AI integration difficult and costly.

AI Innovation Through Hackathons

Flexport utilizes hackathons as a primary driver for AI innovation. The company has observed a dramatic shift, with approximately 90% of recent hackathon projects focusing on LLMs, compared to a much smaller percentage 18 months prior. This bottom-up innovation often leads to ideas that leadership might not have conceived independently.

The Shift from "Manager Mode" to Directive Leadership

Peterson reflects on his own evolution as a leader, moving from a hands-off "manager mode" to a more directive approach. However, he recognizes the value of empowering employees to innovate, particularly through hackathons, and is now re-integrating space for bottom-up innovation into the product roadmap.

Empowering Non-Engineers with AI Skills

Flexport has launched a formal 90-day program for non-engineers to learn AI skills, including coding and application development. The goal is to make these employees "10 times more productive," enabling them to automate repetitive tasks within their roles. This program, initially piloted in Amsterdam, is being rolled out globally.

Impactful AI Projects

Customer-Facing Features:

Natural Language Data Querying: A hackathon project evolved into a feature allowing customers to ask questions in natural language to generate reports and insights from their supply chain data, reducing account management time by 25%.

Internal Operational Automation:

AI for Ocean Freight Planning: A machine learning model optimizes container placement on ships by considering contracts, prices, and sailing schedules. This has resulted in a 2% reduction in ocean freight spend and a 20% improvement in transit times. The AI's ability to parse unstructured data from shipping companies was a key enabler.
Container Re-allocation: The AI system can identify and re-allocate cancelled containers to earlier departing ships, a task too complex and time-sensitive for humans.
Warehouse Address Verification and Appointment Setting: An LLM agent verifies warehouse addresses and schedules delivery appointments via email and voice calls, especially for less frequent delivery locations. This reduces costly errors and missed deliveries.
Customer Sentiment Analysis: LLMs are trained to detect customer sentiment in messaging, triggering automatic escalations to managers for potentially unhappy customers.

Automation Progress and Future Outlook

Flexport has automated 20% of its work by the beginning of the year and aims to reach 50% by year-end. They anticipate reaching 80-95% automation with current AI capabilities, with further increases as LLMs advance.

Economic Implications of AI in Logistics

The labor cost in freight forwarding represents about 10% of the end cost for importers and exporters. Full AI rollout is expected to make ocean freight 10% cheaper. Flexport's goal is to reduce overall shipping prices by 8-9% over the next few years, potentially stimulating global trade. Peterson views AI's societal role as increasing GDP and making goods more accessible, rather than solely focusing on job displacement.

Philosophical and Societal Reflections on AI

Peterson draws parallels between the current AI revolution and the Axial Age, a period of profound societal change driven by impersonal transaction systems. He suggests that just as the Axial Age saw the emergence of major prophets to guide humanity, the current era may require new philosophical frameworks to navigate the implications of AI. He also counters the fear of job loss by emphasizing human nature's inherent desire for more and the role of companies in delivering goods and services efficiently.

The Role of Humans in the Loop

In highly regulated industries like fintech and customs brokerage, human oversight remains crucial. Peterson discusses the concept of "liability sinks" where humans approve AI-generated decisions. He envisions a future where AI forms the core of companies, with humans acting as decision-makers or relationship holders, particularly in serving human needs.

Starting Flexport Today

If starting Flexport today, Peterson would maintain the core principle of not being a pure tech company, emphasizing the importance of human interaction and problem-solving, even for unusual requests like needing a crane for unloading. He stresses that the "tail of things" that are difficult to automate should not be ignored.

Advice on Raising Capital

Peterson advises founders to focus on price per share and control of their company rather than just ownership percentage. He cautions against using capital as a default solution for problems, warning against bloating the company with excessive hiring. His advice for founders raising large rounds is to implement a hiring freeze for 90 days to ensure problems are solved through efficiency rather than just headcount.

Flexport's Vision for 2035

Flexport aims to be a global utility for shipping anything, anywhere, by any means, and in any quantity, accessible via code, APIs, or voice. The vision is for logistics to be as seamless as the electrical grid, allowing brands to focus on product development and customer engagement. By 2035, Flexport plans to have a presence in every country and cover 95% of global container trade with its own operations by 2028. This expansion represents a significant evolution from their initial vision of simply handling customs brokerage.