Uncapped with Jack AltmanAgents in the Enterprise | Aaron Levie, CEO of Box | Ep. 3
At a glance
WHAT IT’S REALLY ABOUT
Aaron Levie on enterprise AI agents, pricing, and platform neutrality
- Levie frames AI as the long-awaited way to activate the massive, underused value locked in enterprise documents—contracts, financials, marketing assets, and HR records—most of which sit idle after creation and storage.
- He describes a near-term future where companies create millions of specialized, content-centric agents (legal, procurement, marketing, finance) that operate inside Box and increasingly coordinate across systems like Salesforce, ServiceNow, and Slack.
- The conversation covers why incumbents may still lose (asleep-at-the-wheel execution gaps, innovator’s dilemma around seat-based pricing, and net-new AI use cases) and why agent pricing should ultimately compress from “labor-comped” to software-like margins.
- Levie also reflects on Box’s early architectural choices (multi-tenant SaaS, no on-prem) as a strategic advantage in the AI era, and briefly discusses U.S. tech politics, California’s governance failures (especially housing), and cautious optimism about pro-innovation tech policy momentum.
IDEAS WORTH REMEMBERING
5 ideasEnterprise data is valuable but mostly sits unusable after creation.
Levie argues the typical file lifecycle is “hot” briefly, then archived for compliance or retrieval—without being queried for insights that could improve sales, onboarding, product discovery, or decision-making.
Box’s AI strategy is to turn content repositories into agent-driven workflow engines.
He envisions (and says Box is already enabling via AI Studio primitives) customers creating agents that review contracts, process invoices, extract metadata from assets, and generate domain-specific analyses directly against enterprise content.
The real agent future requires coordination across many SaaS systems, not just one platform.
Work happens across Salesforce, ServiceNow, Workday, Slack, and more; Levie expects agents will need standards/protocols to call each other and combine data from multiple systems to complete tasks and produce full-fidelity reports.
Startups can win even when incumbents have data and integrations.
Levie lists three durable startup wedges: incumbents that don’t move fast, incumbents blocked by business-model “innovator’s dilemma” (seat revenue threatened by automation), and net-new AI categories where no traditional incumbent exists.
Many AI products are net-new spend before they are cost-savers.
Using AI coding tools as an example, he claims much current adoption is additive—teams swipe a credit card to augment output rather than replacing headcount—creating new budget lines instead of merely reallocating old ones.
WORDS WORTH SAVING
5 quotesMost of it is underutilized... you create one, you share it, you collaborate, and then it goes somewhere, and you never see it again.
— Aaron Levie
You will have... millions and millions of those kinds of agents that get created. They'll help you automate the work that you do with content in every field.
— Aaron Levie
Innovator's dilemma... it's a business model-oriented issue... nobody in the management team wants to have less revenue the next year.
— Aaron Levie
If I had to bet... does AI remain comped at labor... I'm gonna bet on... infrastructure cost plus software and some margin.
— Aaron Levie
Now with AI, that 20-person company could have, like, 10 AI lawyers and 10 AI SDRs... that's the big TAM expansion.
— Aaron Levie
High quality AI-generated summary created from speaker-labeled transcript.
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome