Jul 14, 2025 · 59m · a16z
Aaron Levie on AI's Enterprise Adoption
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
In this episode of the a16z Podcast, Box CEO Aaron Levie and host Martin Casado explore how enterprise AI is reshaping software architecture, corporate productivity, and market dynamics faster than the cloud transition. They discuss SaaS incumbent advantages, AI-native startup opportunities, the evolution of software engineering, and the macroeconomic impact of AI agents.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
The host as informed peer 4.2 Guest teaching 4.1 Guest disagreement 1.4 The host pushing back 2.2
speaking balance: gold is the host, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 25:39 Aaron Defends Domain IP Against CRUD Label
Aaron rejects Martin's assertion that vertical SaaS is merely trivial CRUD software, insisting that domain expertise like clinical trial management is the core intellectual property.
Hardest push from the host ▶ 25:26 Martin Challenges Vertical SaaS as Trivial CRUDMartin directly challenges Aaron's optimism on vertical SaaS durability by arguing that most vertical software applications are technologically trivial CRUD layers.
Biggest teaching moment ▶ 36:32 Aaron's Multi-Trillion Dollar Headcount MathAaron reframes Martin's zero-sum budget query by demonstrating that a tiny fraction of the $5-6 trillion knowledge worker wage pool completely dwarfs total enterprise software spend.
The host holds their own ▶ 38:00 Martin Cites CS PhD Credentials on Programming LanguagesMartin establishes host expertise by drawing on his 30 years as a programmer and Computer Science PhD to explain why formal programming languages will not disappear.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The host as informed peer | Guest teaching | Guest disagreement | The host pushing back | Why |
|---|---|---|---|---|---|---|
| Welcome, Introductions, and Box's Early "Fat Startup" History | 3 | 4 | 1 | 1 | Martin sets up the interview by asking about the transition from prosumer GenAI to enterprise adoption. Aaron provides extensive historical context comparing early cloud adoption hesitancy with today's immediate AI buy-in across enterprise leadership. | |
| Incumbents vs. Startups, Business Models, and Unlocking New Categories | 5 | 4 | 2 | 4 | Martin probes whether AI requires rewriting the full software stack or acts as an API consumption layer, introducing nuances around COGS and usage-based pricing models. Aaron argues incumbent SaaS platforms enjoy sustaining innovation advantages while new AI-native categories unlock massive TAM expansions. | |
| Vertical SaaS Spend, Software Disruption, and AI-First Organizational Design | 4 | 3 | 1 | 1 | Martin highlights how software engineering itself is being disrupted for the first time in thirty years. Aaron explains how vertical categories like legal and healthcare represent uncaptured software spend and outlines Box's internal strategy of leaning into AI-first productivity. | |
| Box's Core Mission and AI as the Unlock for Unstructured Enterprise Data | 1 | 3 | 0 | 0 | Martin invites Aaron to explain Box's core product offerings for context. Aaron delivers a monologue on how AI serves as the ultimate unlock for converting forgotten unstructured enterprise data into automated workflows. | |
| The Bespoke Software Fallacy, Domain Expertise, and the Persistence of GUIs | 6 | 6 | 5 | 5 | Martin pushes a provocative thesis that AI will make bespoke homebrew software trivial and challenges vertical SaaS as mostly simple CRUD applications. Aaron firmly disagrees, explaining that deep domain expertise and workflow operationalization—not tech complexity—are the true IP of vertical software. | |
| AI in Executive Decision-Making, Strategic Planning, and the Bezos Essay Model | 3 | 3 | 1 | 2 | Martin shares how board members are using AI for discussion fodder and provocative insights during executive meetings. Aaron describes using Box AI to anticipate analyst questions on earnings calls and ponders AI-generated deep research for Bezos-style strategic memos. | |
| Enterprise Budget Dynamics: AI Software Costs vs. Multi-Trillion Headcount | 5 | 6 | 2 | 3 | Martin asks if enterprise AI software budgets are zero-sum allocations carved out of existing IT spend. Aaron reframes the math, showing that multi-trillion-dollar knowledge worker headcount budgets easily absorb software tool costs within minor annual salary or hiring adjustments. | |
| The Evolution of AI Coding, "Vibe Coding" Risks, and Enterprise Productivity | 7 | 4 | 1 | 3 | Aaron asks Martin for his view on AI coding, prompting Martin to draw on his CS PhD background to argue why formal programming languages will endure over pure natural language prompts. Aaron elaborates on the new work paradigm where humans act as auditors fixing AI-generated mistakes. | |
| Measuring AI Impact: Internal Velocity, Economic Metrics, and Future Workflows | 4 | 4 | 0 | 1 | Martin asks for the best economic metrics to track AI's macroeconomic pace. Aaron shares Box's internal directive to use AI to drive organizational velocity and capacity rather than pure headcount reduction, predicting long-term societal productivity gains. |
Statements from this episode (25)
Levie: Enterprise AI adoption will outpace the cloud computing transition
“AI is going to take over the enterprise. We know this is going to happen, and it needs to happen to us faster than it happens to our competitors, which is a totally different dynamic than we saw with cloud.”
Levie: Box scaled aggressively using Ben Horowitz's 'Fat Startup' playbook
“And we let's just say we very much took that to heart. And we basically like deployed every single lesson. Which was, like, the name of the game is you get big fast, you scale aggressively, and that was a very important kind of period in our company's journey.”
Levie: Enterprise AI adoption speed is limited by human workflow changes
“And that, that's something where I do think We have to be prepared for this as many years. It's about the speed at which humans can change their workflows as opposed to how kind of quickly the technology can just, you know, sort of evolve in advance.”
Levie: Enterprise buy-in for AI is 5x higher than early cloud
“So what's really interesting is I think the level of buy-in you have now in the enterprise is like five times, you know, greater than we had in the early days of cloud.”
Levie: Goldman Sachs uses AI to draft S-1s in minutes
“David Solomon at Goldman Sachs, Has given this anecdote of they can write now an SEC filing or an S one for an IPO in like a few minutes that used to take a number of analysts, you know, a few days.”
Levie: AI agents act as sustaining innovation for SaaS incumbents
“Right now with AI, everything kind of looks like a sustaining innovation if you're an incumbent, which is like, instead of a user pressing the buttons in the application, let's have an agent run through the API and operate as if they were that user. And so all…”
Levie: SaaS per-seat pricing will persist until humans are fully removed
“Well, until you, until the human literally is not a seat on the system, I think you don't remove the end user license as a component.”
Levie: Founder-led SaaS incumbents pivot to AI faster than on-prem predecessors
“Another idiosyncrasy is I would say like on the margin, you tend to have founders still leading the SaaS companies. And so we didn't really have that in the on-prem world. And so you like Siebel already had three CEOs later and PeopleSoft already had multiple …”
Levie: AI agents unlock startup opportunities in non-digitized sectors
“For startups in the AI world, the equivalent of that is I think there's a lot of categories now where there is no actually software incumbent in that category where AI agents all of a sudden let you go build software for that category. Legal, healthcare, educa…”
Levie: Individual contributor roles will shift to orchestrating AI agents
“And it's like, okay, your job is now orchestration, you know, integration of work, planning, task management, reviewing, auditing, and that will, you know, that will radically change work.”
Levie: Companies shouldn't lock in operational transformations for current AI
“Interestingly, there's probably, it probably behooves us to not over rotate on transforming yet internally for any given company, simply because the technology is changing so fast that like you probably wouldn't want to snap the line right now, run your whole …”
Levie: Over 90% of workers do not want to prompt AI
“I think if you ask like 90%, if you ask the world population, you know, you probably find that 90 plus percent just don't care enough. They just don't like, they don't care about the tabs on their software and the modules on their dashboard. Like it's like the…”
Levie: AI will trigger a 10x surge in custom software creation
“And so I could underwrite a 10 X growth of the amount of custom software that gets written and the fact that these core systems don't go away because there's just actually going to be way more software in the world that gets created.”
Levie: Vertical AI agents will outperform horizontal models in specialized workflows
“Now, that translates fine to agents, but I still would then bet on that, that vertical player doing that, as opposed to somebody prompts their way into ChatGPT to build a FDA compliance agent. I would still largely bet on, you know, complianceagent.ai to do th…”
Levie: Graphical user interfaces will remain essential alongside AI agents
“The theory of, like, the full abstraction away from the interface, and it's all an API call, I don't think that happens, and so, ironically, probably what will happen is in a couple years from now, we will see agents, like, rebuild entire web pages and dashboa…”
Levie: AI deep research queries have replaced chief of staff tasks
“The kind of thing that three years ago I might sort of lob over to the chief of staff and say, hey, can you like go research, like the pricing strategy of this ecosystem or something? That's just a deep research query now. And then I'll wake up and I'll look a…”
Levie: AI will invent new work rather than just automate tasks
“What we do too many times with an AI is we like look at today's way of working and we're just like, AI will come in and take 30% of that. And it's like, no, no, no, we'll just do totally different things with AI.”
Levie: Enterprise AI coding tools cost just 1% of an engineer's salary
“Let's say your most aggressive cursor usage or something is like a thousand bucks a year, 2000 bucks a year. So you're at like one percent salary maybe.”
Casado: AI will not eliminate formal programming languages
“I just don't see a world when, where you get rid of formal programming languages just because they arose out of natural languages for a reason. Like we started with English and then we made programming languages so that we could formally describe stuff. And so…”
Levie: Incoming software engineers will be unable to code without AI
“It's going to change the nature of the incoming class of engineers that you hire. Like they will literally not be able to code without AI assistant, assisting them.”
Casado: I code most nights using Cursor because it's enjoyable
“Many of them, I, like I code, you know most nights with cursor just because it's really enjoyable.”
Levie: AI produces million-dollar marketing videos for hundreds in token costs
“You can now put together an otherwise million dollar marketing video for a couple hundred bucks of tokens.”
Levie: Box uses AI to boost operational capacity, not cut costs
“Internally, we've actually explicitly taken the stance that we want to use AI to increase the capacity and capability of the company, so just do more.”