Opportunities in logistics tech
What AI means for the next wave of logistics startups & more
The last few years in logistics have been a rollercoaster. COVID created supply chain crunches and chaos as prices soared to never-before-seen heights. Carriers and brokers experienced record profits while shippers prayed for a reversion to the mean. In tech, suddenly everyone had a supply chain thesis and investment poured into the sector.
Today, carriers and brokers are feeling the pain, including prominent VC backed companies. The word bloodbath is even being thrown which is never a good sign.
So, what does one make of this chaos? Is logistics tech a zero-interest rate phenomenon?
My view is resoundingly to the contrary. Despite the recent news cycle, there’s still plenty of reasons to be long-term bullish on logistics tech.
The last wave of logistics tech: A refresh
This post is not intended to be a 101-style write-up on logistics, although I am happy to share something to that effect if it’d be valuable. That said, if you’re already plugged into the logistics tech ecosystem, feel free to skip this section as it may be a bit repetitive.
Much of the success of the last wave of logistics tech came from one of two approaches — build a tech-first services company to out-compete legacy incumbents (e.g., Flexport, Convoy, Uber Freight) or develop foundational infrastructure for the ecosystem (e.g., Project44, Samsara, Motive, etc.)
Tech-first services
When an industry is highly fragmented and low NPS there’s often an opportunity to go beyond SaaS and build a vertically integrated company to out-compete incumbents. Vertically integrated logistics companies sacrifice margins for a greater share of total dollars in the industry. Further, you eliminate some of the headaches of trying to sell software to legacy organizations with archaic tech stacks and cemented ways of working. You can build from a clean slate without having to worry about industry readiness for adoption. This was the path followed by Flexport, Convoy, and Uber Freight. The challenge of this model, as we’re now seeing, is you’re more exposed to the forces of the markets which, in logistics, are extremely volatile. The highs are high and the lows are very low.
Foundational Infrastructure
One challenge with selling software to industries that exist in the physical world is that there’s a lot of critical data that exists out there in the “real world.” As such, there’s a certain order of operations required for software to effectively penetrate an industry like logistics - in this case, step one is connecting to the physical infrastructure that underpins the industry. Telematics was a major winner here and has enabled a number of other use cases. For example, without telematics, it would have been incredibly hard for the advances in fuel cards (AtoB, Coast, etc.) because there would be no way to validate a purchase location which is a major input into fraud prevention. It’s awfully hard to build software for an industry focused on the movement of goods if you don’t have a way to track where the goods actually are. Project44, Samsara, and Motive are examples of companies that laid this foundation.
What’s next
We’re now in a world where a few things are true:
The industry is still huge. Really huge. The logistics market is often measured as a percent of GDP and is ~$10T globally.
The industry still has a relatively low degree of software penetration. McLeod, a legacy TMS vendor, said in September of 2022 that only ~27% of their customers used the cloud version of their product.
The industry is increasingly connected due to telematics solutions, digital brokers, and back-office software, like SmartHop. This provides foundational building blocks for new products as well as new distribution channels (e.g., AtoB’s partnership with Uber Freight) for more capital-efficient growth.
There’s an increasing number of “logistics enabled” entrepreneurs who are well-versed in the dynamics of the logistics industry through experiences at companies like Flexport, Uber Freight, and others.
AI creates a paradigm shift in the product experiences that can be delivered. LLMs are uniquely suited to unlock the massive amounts of unstructured data in the industry and automate key workflows.
With all of these foundations in mind, below outlines the opportunities I’m most excited about right now.
Automating manual workflows
Much of the logistics world still lives in paper documents, PDFs, and/or is locked up within legacy systems with lackluster integrations.
There are two steps here: getting the data and then acting on it.
Traditionally, the data access step is done through a combination of direct integrations, carrier APIs, CSV uploads, and/or EDI connections. Companies like One Schema (CSV), Terminal (APIs), and Orderful (EDI) are all streamlining this data exchange process.
With advances in AI, some companies are taking this a step further by extracting unstructured data from sources like invoices (e.g., Loop, Narrative), emails (e.g., Vooma, Linc AI), and phone calls (e.g., Fleetworks).
Having access to this data enables a number of multi-touch, manual workflows to be automated, such as:
Finance: Logistics invoices are extremely complex. There are often many accessorial line items that vary based on the carrier and even within carriers depending on the negotiated rates. A shocking ~20% of invoices are incorrect leading to the use of offshore labor to manually reconcile invoices. This is a slam dunk use case for LLMs, which companies like Loop are focusing on.
Customs: Similar to invoices, customs is a clear use case for LLMs. It’s complex, time-consuming, and grounded in text. Companies like Expedock and Raft are building in this space.
Order entry: Much of the logistics industry still does business over email or on the phone. The beauty of AI is that it allows companies to continue to do business with their counterparties through the same systems (email, phone, etc.) while seeing all of the efficiency benefits of a more automated process. Companies like Vooma, Linc and Fleetworks are building here.
In addition to startup activity, this is an area that larger players, like Flexport, are already investing in. Below is a quote from Flexport’s CEO, Ryan Petersen regarding their use of AI:
“Labor cost in coordinating freight is ~10% of the cost of international shipping. AI will make almost everything you buy cheaper.
Thanks to our new GPT-4 based copilot that Flexport ops tech team rolled out this week, a task that used to take operators 30 minutes can now be done in 20 seconds with a single prompt.
A huge number of similar tasks in freight forwarding will be susceptible to similar AI-based automation. We plan to knock them all down one by one in the months ahead.”
Optimizing operations
Logistics is full of multi-variable optimizations that touch everything from pricing to how to best fill a truck. There’s a lot of math and machine learning that larger players already use to drive meaningful efficiencies. Beyond the algorithm, there’s also a visibility problem. Most companies in the supply chain see a mere fraction of overall activity in the market which makes it hard to understand the ‘true’ market price. This visibility problem creates a clear opportunity for a platform and one that would likely benefit from compounding advantages over time.
Some startups building in this area include Greenscreens.ai (Pricing), Goodship (Pricing), Aircon (shared air cargo), Iso (Performance data), and many others.
Note: The security certificate for Flexport’s engineering blog expired a few days ago so some links are not working at the time of this posting.
Offering financial services
Logistics is complicated and with that complexity comes a complicated web of financial services whether it be B2B payments, fuel payments, debt financing, or insurance. This is an area where getting distribution right can be a major unlock, whether that is partnering with another logistics tech company, like Uber Freight, or attaching fintech to an adjacent product. MVMNT, for example, offers a free Transportation Management System (TMS) in order to monetize via Factoring. Lula offers a pay-per-mile insurance product and will launch an embedded insurance product in the future.
While the idea of fintech in logistics is less greenfield than AI-driven opportunities, there’s still a huge market for disrupting incumbents and increasing the penetration of financial products. For example:
In fuel cards, incumbents WEX and Fleetcor (founded in 1983 and 2000, respectively) have a combined ~$25B market cap. Suffice to say, the product experience is what you’d expect from two companies founded over 20 years ago.
In insurance, there’s still a huge coverage gap (the delta between what shipments should be insured and what shipments are actually insured). Loadsure, an Insurtech MGA, estimates that 60% of shipments today are under or uninsured.
Some startups building in this area include MVMNT (TMS w/factoring), Lula (Insurance), Nirvana (Insurance), Breeze (Insurance), AtoB (Fuel cards), and Onramp (fuel cards).
Closing visibility gaps
While visibility has improved dramatically thanks to Samsara, Motive, Project44, and others, there are still gaps. This can be an incredibly costly problem - losing track of your assets can result in demurrage or detention charges. These fees continue to climb to astronomical levels year over year.
In particular, two notable visibility problems today are freight rail and the yard.
In freight rail, demurrage charges continue to climb. To set the scene, rail is one of the underpinnings of the North American logistics ecosystem generating ~$90B in revenue and moving ~1.5T revenue-ton miles (one ton of freight moving one mile) of freight per year. The Class I US railroads were on track to collect $1.6B in demurrage charges in 2021 which was the highest of all time. Telegraph is trying to address this pain point with a platform purpose-built for freight rail.
Another visibility black hole is the yard. In this case, it’s the yard operators who lack visibility into the assets that have entered their yards. As a result, visibility over operational KPIs, like wait times, is non-existent. Further, yard operators in certain geographies need precise visibility into the number of trucks entering their yard to comply with environmental regulations. Terminal is building a solution in this space.
Logistics tech landscape
The market map below outlines the segments covered above. To be clear, this is not an exhaustive list of companies in logistics tech.
Common traps for logistics startups
Picking an ICP: Not too big, not too small
Even within a single market segment, like Over-The-Road transportation, there is significant variation in the size of customers (e.g., from mom-and-pop to public company) and sub-segments of customers (e.g., FTL, LTL, Flatbed, etc.). This creates a perfect storm for a product offering that tries to be everything for everyone and ends up falling flat. Conversely, going too narrow (e.g., TMS for Flatbed in Mexico) can result in a business model that is not venture scale.
Putting in the time early to consider which segment(s) can support a venture scale outcome and what wedge provides the best long-term opportunity is critical.
Implementation purgatory
Technology bandwidth can be tight for logistics companies and a multi-month implementation cycle is not unheard of. Minimizing time to value is critical. In service of this, good sales qualification processes and Customer Success teams are existential. Never-active churn, which is the churn attributed to customers who never fully onboarded to the product, can also be a challenge. Startups should closely monitor the gap between booked and recognized revenue to ensure this doesn’t spiral into a larger issue.
Failing to build for multiple market cycles
Logistics markets can be volatile and down markets often result in significant challenges for market participants. Research by Convoy (ironic given they recently shut down) found that there have been 12 freight recessions since 1972 which is more than ~2x the amount of recessions in the overall economy. The hardest hit segment in a freight recession is often the owner-operator segment of trucking (i.e. SMBs) as they rely heavily on the spot market.
Logistics startups need to be aware of the climate they’re building in and be prepared for when the music inevitably stops. Building an efficient business and taking swift action when a recession inevitably does come can help weather the storm.
Final thoughts
I am super excited about the next wave of startups in logistics.
The last wave of great logistics startups has laid the foundation — both in terms of technology and inspiring future founders — for an explosion of growth in the space. Topping it off, AI has fundamentally changed the types of product experiences that can be delivered. This has unlocked a whole set of use cases that weren’t possible as little as 12 months ago. Who knows what will happen in the next 12 months…
If you’re building something in logistics, I’d love to meet you! You can reach me at matt@renegadepartners.com