Why AI Is Reinventing How Businesses Buy Everything
a16z · 2026-10-02 · 59 min
https://www.youtube.com/watch?v=OTQ-lFsq7zAVideo summary
Lio CEO Vladimir Keil says AI agents can manage procurement end to end, where one missed aircraft-part email could cost hundreds of millions.
a16z’s Seema Amble and Elena Burger speak with Lio CEO Vladimir Keil about how AI startups can outperform incumbent software by taking on work beyond a system of record. An ERP might show an $8,000 aluminum purchase, while hiding 500 emails, 20 spreadsheets, supplier pushback, and weeks of engineering analysis. Keil says Lio’s multi-agent system handles sourcing, RFQs, negotiation, orders, shipment tracking, and invoices, with human review scaled to risk; enterprise customers can build trust over 10,000 to 100,000 negotiations. In aircraft and data-center projects, a missed supplier delay email can cause hundreds of millions in damage. The speakers also discuss incumbent incentives, vertical AI moats built through customer trust and end-to-end work, and the prospect of agents coordinating both buyers and suppliers. Keil says procurement remains frustrating despite a crowded software market, but even 1% savings can match the P&L impact of 10% more sales.
Chapters
- 0:00Why AI startups still beat incumbents: owning end-to-end work beyond systems of record
- 4:19The hidden work behind an $8K line item: supplier pushback, 500 emails, and cost engineering
- 6:18Retrieval, process, policy, principal: from outage lookup to relationship-based compensation
- 9:27The incumbent's internal conflict: Salesforce Agentforce, product incentives, and Lio’s 80% exception work
- 14:48What procurement actually looks like: cross-functional experts, human review, and overlooked purchases under $50K
- 21:13Procurement at Boeing-scale: thousands of suppliers, costly delays, and coordinated multi-agent work
- 28:03A bolt order, end-to-end: Lio automates sourcing, RFQs, and order tracking
- 32:52A bolt order, end-to-end: negotiation prep, live oil-cost analysis, and should-cost models
- 37:57What a durable vertical AI company looks like: owning workflows beyond an eight-hour prototype
- 44:46When both sides deploy agents: procurement can bring suppliers onto AI workflows
- 50:19When both sides deploy agents: Lio’s 85% engineering team and 100-plus procurement leaders
- 54:14When both sides deploy agents: procurement’s outdated workflows and trillion-dollar potential
This is a Tier 1 public summary
Whether the chapter key points, section summaries and mind map are public is up to the person who shared it. Want the full analysis?Submit one yourself.
More from this channel
Why One Internet Pioneer Thinks the Original Model Brokea16zDoxxNet founder Barrett Lyon is building a 196-domain mesh for private peer-to-peer calls, chats, and 20 GB file transfers.
Why Specialized AI Could Beat The God Modela16zOpenRouter’s Alex Atallah and Replit’s Amjad Masad argue specialized AI can outperform a single god model, with fusion reaching frontier quality at 40–50% of the cost.
The Current State of Consumer AIa16za16z finds only 4.5% of US consumers pay for AI, while OpenAI’s ads reach a $1 billion annual run rate.
How Cursor Built One of AI’s Fastest-Growing Companiesa16zCursor won users before building models, defied Microsoft’s Copilot, and reached over 50% of Fortune 500 companies.
Why Top Founders Are Racing Into AI Infrastructurea16za16z’s Machine Age Fund targets AI infrastructure as GPU supply sells out through 2028 and memory demand needs three years of capacity.
Related analyses
Why AI Agents Need Context | Deep Dives with a16za16z Deep DivesFivetran CEO George Fraser says AI agents need centralized data, while warning AI-native rivals could challenge SaaS incumbents.
How AI Will Rewire Consumer Internet | Deep Dives with a16za16z Deep DivesJosh Elman joins a16z after Apple, arguing consumer AI must deliver 10x-better experiences and lasting retention.
Cybersecurity in the Agentic Era | Deep Dives with a16za16z Deep Divesa16z, Neo, and Cotool warn that 50% of enterprise apps could be agentic by year-end, making traditional security defenses unreliable.
Why AI Demand Is Outrunning Compute Supplya16zGavin Baker says AI demand may outstrip compute through 2028, while Nvidia, open-source models, and orbital data centers all stand to benefit.
Inside the Race to Measure Frontier Intelligencea16zVals CEO Rayan Krishnan says Llama 4 underperformed private tests, while AI token costs can exceed salaries 10x.