Sep 5, 2025 · 42m · american-optimist
How AI Agents Are Transforming Logistics · Joe Lonsdale
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Host Joe Lonsdale interviews serial entrepreneur Harish Abbott, founder of Augment and former CEO of Deliverr, exploring his journey from Amazon to building billion-dollar logistics ventures and how AI agents are eliminating supply chain waste to transform global trade.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Joe holds 20.4% of the talking time here. How this is scored →
Joe as informed peer 4.0 Guest teaching 3.8 Guest disagreement 0.0 Joe pushing back 0.2
speaking balance: gold is Joe, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 20:06 Harish dismantles the centralized freight platform thesis
Harish contrarily rejects the conventional Silicon Valley assumption that digital marketplaces can easily unify trucking, detailing how asymmetric pricing risk forces truckers to hide critical data.
Hardest push from Joe ▶ 15:04 Joe presses on the messy Shopify-Flexport spin-outJoe challenges the smooth corporate narrative by explicitly probing the internal complications and messy dynamics of Shopify divesting Deliverr to Flexport.
Biggest teaching moment ▶ 29:35 Harish breaks down invisible macro waste in freightHarish educates Joe on the structural math behind hundreds of billions of dollars wasted when missed dock appointments cascade into overnight truck layovers and idle warehouse shifts.
Joe holds their own ▶ 37:58 Joe dissects H-1B controversies between startups and IT consultanciesJoe demonstrates deep industry insight by articulating the policy distinction between hyper-growth technical builders and low-cost outsourcing body shops like Infosys.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Joe as informed peer | Guest teaching | Guest disagreement | Joe pushing back | Why |
|---|---|---|---|---|---|---|
| Show Opening Title Sequence | 3 | 1 | 0 | 0 | Joe introduces Harish Abbott, noting their investor-founder history across multiple companies, and sets up a discussion on immigrant entrepreneurship. Harish reflects on his upbringing in India, studying graph theory, and coming to America for its meritocratic culture. The exchange is warm, highly aligned, and introductory. | |
| Early Career at Amazon & Lessons in Obsessive Leadership | 3 | 5 | 0 | 0 | Harish shares formative lessons from his early days building Amazon's fulfillment systems, highlighting Bezos's customer obsession and Jeff Wilkie's operational reviews. Joe interjects with affirmations about exception handling and proactive business disruption. Harish provides deep operational detail on managing 150 daily metrics across warehouses. | |
| Operational Principles: Owner Mindset & Writing Culture | 4 | 3 | 0 | 0 | Harish outlines key Amazon operating principles, focusing on owner mindset and the mandatory multi-page memo culture before meetings. Joe validates the practice by sharing how rigorous written memos alter investment decisions at his own firm. Both agree on the discipline writing enforces against blind spots. | |
| Building Deliverr: The AWS for Physical Logistics | 5 | 5 | 0 | 0 | Harish explains the genesis of Deliverr as an AWS-like programmable logistics layer for non-Amazon merchants. Joe actively engages using CDN analogies and warehouse density economics. Harish details how multi-merchant predictive data pooling enabled one-to-two-day shipping for smaller businesses. | |
| Shopify Acquisition & The Spin-Out to Flexport | 4 | 4 | 0 | 1 | Joe asks directly about the friction behind Shopify acquiring Deliverr and subsequently offloading logistics assets to Flexport. Harish provides a candid structural explanation of the margin clash between high-margin pure software and low-margin physical fulfillment. The tone remains analytical and cooperative. | |
| Structural Fragmentation in Freight & The AI Imperative | 4 | 6 | 0 | 0 | Harish explains why he returned to entrepreneurship, citing LLM reasoning capabilities and context windows as the unlock for fragmented supply chains. Joe asks why centralized platforms like Uber Freight and Convoy struggled to fix the market. Harish educates Joe on the structural game theory and low-trust information withholding among independent truckers. | |
| Augment & Augie: The Multimodal AI Logistics Teammate | 3 | 4 | 0 | 0 | Harish describes Augment's core product, an AI teammate named Augie that handles multimodal communication across email, phone, WhatsApp, and Telegram. Joe inquires about scale and workflow automation. Harish shares that Augie is already deployed across $25 billion of freight under management. | |
| Managing AI Mistakes, Adversarial Attacks, and Edge Cases | 4 | 4 | 0 | 0 | Joe presses on failure modes, hallucination risks, and adversarial hacking attempts against AI agents managing financial transactions. Harish explains their guardrail architecture, categorizing low versus high-consequence mistakes, and describes how internal agents red-team Augie. He shares a real-world example of an accidental automated late notification sent to a client. | |
| The Macro Economic Opportunity: Eliminating Supply Chain Waste | 3 | 6 | 0 | 0 | Joe asks about the total economic opportunity and whether current model intelligence is sufficient. Harish lays out the macro dynamics, pointing out that 10% of logistics value is lost to asynchronous scheduling delays and overnight truck layovers. He quantifies this as a $300 billion domestic and $1 trillion global efficiency unlock. | |
| Recruiting Top Talent & Building Engineering Culture | 5 | 2 | 0 | 0 | Harish discusses recruiting engineering talent and expanding to Waterloo and Toronto. Joe shares his own operational experiences at Palantir, comparing Midwest engineering retention and loyalty favorably over coastal tech culture. The dialogue is a collaborative founder-to-founder exchange on talent architecture. | |
| Immigration, Meritocracy, and High-Skilled Talent Debates | 6 | 2 | 0 | 1 | Joe brings up the contentious H-1B visa debate, asking Harish how high-growth startups navigate immigration quotas. Harish emphasizes maintaining high meritocratic standards regardless of origin. Joe reframes the broader political debate, sharply distinguishing high-output technical builders from body-shop wage-arbitrage firms. |
Statements from this episode (19)
Abbott: No country matches America's trust in immigrant founders and embrace of failure
“Well, I mean, I don't think there is any country in the world like America, where folks from all backgrounds, from all countries can come and can be trusted to go build businesses,
Endeavors, social endeavors and just embrace of failure in, in order to go buil…”
Abbott: Bezos priced Kindle at $9.99 to cannibalize Amazon's book business
“Jeff sort of incubated the Kindle project, which, if you look back, could be like, oh, you're
Sort of cannibalizing the biggest business you have, because now you're going to sell every single book for nine dollar 99 cents.
Yeah.
When the average prices of …”
Abbott: E-Commerce Remains Sub-20% of Total Commerce
“E-commerce at that time, and I think still is, like, sub-twenty percent of all commerce at that time, it was close to 12%.”
Abbott: 1-2 day shipping requires a CDN for physical goods
“And really the only way to get to fast shipping is Almost like building a CDN for the physical world.”
Abbott: Shopify powers close to 12% of U.S. e-commerce
“Shopify was the fastest growing, you know, e-commerce platform. It's now, I think, close to 12% of U.S. E-commerce. So pretty meaningful, you know, the power, I think close to 250 to three hundred billion of online commerce now.”
Abbott: Flexport's warehouses are full and adding capacity
“I think the last I heard from Ryan is their warehouses are full. They're adding more capacity. It's a pretty meaningful business now.”
Abbott: Logistics fundamentally is trading information to trade goods
“Like, if you take a step back, Joe, like, logistics in its simplicity is about trading information so you can trade goods.”
Abbott: There are 1 million trucking companies in the US
“So there's one million truck driving companies in America, right?”
Abbott: US has 40,000 to 50,000 freight brokerages
“There's 40, 50,000 trucking brokerages in America.”
Abbott: Augment deploys AI across $25 billion in managed freight
“It is up and running. It is, you know, doing it at scale you know, collectively today you know, across our businesses getting deployed to about twenty five billion of freight under management. So companies collectively managing that much freight and it's growi…”
Abbott: Augment built AI agents solely to red-team and hack Augie
“We built agents that tried to, their only job is to hack Auggie, because, you know, it is out in the wild. It is handling tens of thousands of loads.”
Abbott: Logistics represents $3.2 trillion or almost 10% of US economy
“And in the US, it's 3.2 trillion. Our economy is, I think, 35, 36. It's almost 10% of the economy is logistics.”
Abbott: Logistics industry accounts for roughly $80B in payroll
“There's about eighty billion of payroll in the industry.”
Abbott: Asynchronous communication causes a $300B waste problem in US logistics
“We think there's about at least 10% waste in this industry because of the low bandwidth asynchronous communication nature of trading information. So that's a three hundred billion dollar opportunity in the US, almost a trillion dollar globally, right?”
Abbott: H-1B visas should be fast-tracked for exceptional talent, not routine jobs
“The folks who are coming on H-Bs today, like me, like many other people who came at some point, you know, do end up creating jobs in this country and do end up, you know, helping build businesses in this country. So I'm on the take, which is like, let's not lo…”
Lonsdale: H-1B backlash stems from IT body shops, not elite startups
“I think there's a confusion with the argument with a lot of people who aren't in the tech world where there's, I think there's two types of businesses. One type of business is more like yours where it's just like super fast growth. It's clear how this could be…”
Abbott: US Cannot Compete on Labor Costs, Must Compete on Data Efficiency
“We don't, we'll never have the lowest labor cost, right? Like we can't compete on that, but what we can compete on is efficiency through data.”