Anthony Sutardja on Revolutionizing Logistics with Generative AI

For future installments of the Supply Chain newsletter, subscribe here.

We’ve spent the last year looking at applications of AI in supply chain operations. We’re convinced that AI will give logistics companies super powers, ultimately enabling them to work with more customers and vendors than ever before. The end result should be healthier logistics companies, more liquidity in the logistics industry, and, ultimately, more durable supply chains.

We look at Parade, led by CEO Anthony Sutardja, which has undergone a remarkable transformation since the roll out of its AI product offering. Initially established as a freight brokerage, Parade pivoted its focus towards capacity management software for freight brokers in 2018. In 2023, it released Parade CoDriver, an AI assistant to automate conversations between freight brokerage operations teams and the trucking companies that service them. The result has been a 6x increase in productivity to the average Parade user! I sat down with Anthony to learn more about Parade; how it’s always been an AI-first company, building its initial product based on analysis of carriers’ emails; and how Generative AI is massively scaling the work as well as enabling much richer client relationships.

Why did you start Parade? What did it start as, and what is it today?

We initially ventured into building a freight brokerage, drawing inspiration from experiences at Uber and personal connections in the trucking industry. However, after two years of operation, we realized that rather than competing with freight brokers, our true opportunity lay in empowering them. This realization led us to pivot our focus towards providing tools and solutions that streamline the core business of buying and selling truckload capacity. Our platform assists brokers in finding carriers more efficiently, automating booking processes, and leveraging data to optimize pricing strategies and secure freight contracts.

Who is Parade’s customer today and how does Parade enhance their value proposition?

Our customers primarily consist of freight brokerages, ranging from lower mid-market to mid-market enterprises. Regardless of their size, they all share a common need: to expedite truck-finding processes and improve overall efficiency in their operations.

While some have developed their own tech stacks for digital freight management, we offer a more targeted AI solution that seamlessly integrates with their existing systems. For those with core TMS systems, we provide an overlay that connects them to our platform, facilitating their transition into the digital realm of capacity management.

Our solutions really cater to a broad spectrum of clients, offering tailored approaches for enterprises and mid-market companies alike.

Are your customers able to recognize the value in terms of increasing the number of loads moved per rep in a day and securing more volume? Essentially, if I'm investing in Parade's core offerings, how can I see that it significantly adds value to my business?

Absolutely. At the core, it's about automating booking processes to enable our teams to accomplish more with less effort. Before Parade, the average carrier sales rep at a brokerage handled around 8 to 10 loads per day, per person. With some effort, this could stretch to 15. Pre-AI, or pre-Generative AI, this number might reach 20 to 25 loads per day. However, post-Generative AI, we’re at 50 loads per day, and we don’t know what the ceiling looks like. We’ll continue to push these boundaries of what’s possible.

Brokers are also exploring additional avenues for value delivery, particularly focusing on carrier reuse as a means to enhance gross margins. This involves two key aspects: firstly, optimizing workforce efficiency to achieve more with existing resources, and secondly, maximizing the spread between shipper and carrier rates. By prioritizing carrier reuse, brokers can stabilize rates and consequently improve their average profit margins, leveraging Parade to facilitate this process.

How exactly is Parade using Generative AI to enhance its value proposition? And how did you land on the right use case for Generative AI within Parade?

We've always been an AI-first company. We got our start in natural language processing (NLP) of emails from truckers back in 2019. This allowed us to help them efficiently process and extract crucial information from carriers. It was like a massive data upload for our customers, who often felt like they were searching for a needle in a haystack. When carriers inquired about available trucks in a specific region, we could swiftly understand their needs and respond with suitable matches.

With Generative AI, we've transitioned from direct extraction to engaging in multi-turn conversations, which has significantly enhanced our capabilities. What previously took us a year and a half to build can now be achieved in a weekend with modern tooling. This advancement has opened up new possibilities, particularly in addressing the bi-directional negotiation challenge between brokers and carriers.

While we could surface matches directly to them before, we now understand that true value lies in engaging carriers in meaningful conversations to negotiate and move transactions forward. Generative AI enables us to have extended dialogues with carriers, responding to their inquiries and negotiating prices over multiple turns. It's a capability we've been closely monitoring across our suite of products, as it aligns perfectly with our mission to deliver maximum value to our customers.

We offer a private carrier portal through email, allowing brokers to provide carriers with a bid and book button. While some carriers readily adopt this digital platform, others prefer to communicate through email and phone. That's kind of that last bastion of freight that has not been digitized for us. And so, more specifically for us, it aligned very nicely with the problem areas that we were trying to solve in terms of getting the transaction digitized.

Has that transformed that fundamental value proposition that you initially had before the implementation of Generative AI?

It’s expanded our ceiling. Before, peak digitalization for our customers was probably around 45 to 50% depending on the month. The long tail customers were maybe sitting around 15 to 20% digital. With the introduction of Generative AI, those at the lower end of the spectrum jumped from 10 to 35% digital within just 30 days.

And it isn’t just revolutionizing peak outcomes; it's reshaping our approach to initial value delivery and guiding us toward a more efficient go-to-market strategy. By leveraging Generative AI, we can rapidly introduce a full suite of solutions to our customers, accelerating the adoption process and ensuring they experience the transformative potential of our products from the outset. It’s a great way to get a taste of Parade a lot faster than before.

You said there was a certain number of loads that a rep could move without any digital tooling. With Parade, they were able to move more, but with Parade Generative AI that can continue to increase. Can you share how meaningful that’s been in terms of loads per day, per rep?

It’s been a big jump. We're still in the early stages, about three months into our production rollout, but we recently conducted a case study focusing on one representative at a company. The typical rep would book 8 to 10 loads per day, per person. With Parade Generative AI, the rep at this company was doing 8 loads per hour.

This significant increase in productivity is due to us handling the qualification process for them, enabling faster price action and simultaneous conversations within 10 seconds. Previously, slow responses to calls and emails led to carriers being booked out. This represents a fivefold increase in efficiency.

I think the biggest challenge we're facing during this rollout period over the next year is reevaluating and restructuring work allocation within brokerages. As we progress through the freight booking pipeline, the next crucial step is implementing the connection to track and trace systems and ensuring carriers are using tracking tools effectively. It’s really exciting. And I think the next piece for a lot of these organizations is always thinking through operational considerations.

What were the most significant challenges you faced in developing a market-ready Generative AI product that effectively adds value to your customers? Were there any initial ideas or applications for the technology that you envisioned but ultimately proved unsuccessful during the development process?

One challenge we encountered was focusing too much on cost considerations, which ended up being a waste of time. In hindsight, we should have assumed that costs would naturally decrease, given the increasing adoption of such tools and the expansion of infrastructure. It’s a big assumption, but it is happening and I think it will continue to happen. That was one piece that kind of slowed down our development.

We're experiencing a paradigm shift in how we approach conversations, moving beyond simply progressing from point A to point B. For instance, when a carrier expresses interest in a load, our objective is to transition that conversation into price action. Coaching and framing the AI to achieve this goal has been a fascinating journey.

Engineers play a crucial role in shaping this process, but there's also a new skill set needed within our organization. I like to refer to this as conversational UX. It involves understanding how to design conversations effectively, guiding the AI toward desired outcomes, and ensuring a seamless user experience throughout the interaction. We're investing in individuals who can really help shape and continue to iterate on these conversations effectively due to the vast potential in this area.

Looking at the success of your AI product, what broader implications does it suggest for the applications of AI in logistics software? Based on your experience, where do you think the biggest opportunities are?

I think some of them are already in progress, specifically these general knowledge-based Q&A products. I think that's so general that every single business in logistics can use them. Do you create a custom training solution, or do you opt for off-the-shelf tools? I lean towards the latter, where it's broad enough to simply plug in documents. Then you've got a top-notch brokerage coach for your team with all the SOPs readily available, making hiring new staff as easy as possible. I see this as a promising area for improvement, as I believe many general solutions can already address this need.

I also think there are a lot of opportunities with machine-to-machine translation being accelerated. I think they’re very good at talking code. They're very good at writing code. They're very good at translating not only English to Portuguese but also Node to Python. In freight, there are some archaic formats out there that are very cumbersome to navigate. There’s a huge opportunity for Generative AI to help streamline this integration process, bridging the gap between outdated forms of integration.

Lastly, I believe there's a crucial aspect related to support-level workflow automation. This aligns with our current focus on one of our products. Conversations that facilitate business transactions can largely be automated, as they primarily involve exchanging information.