---
type: x-post
source: https://x.com/levie/status/2061247380897579500
author: Aaron Levie
author_handle: "@levie"
company: Box
captured: 2026-06-01
posted: 2026-06-01
tags:
  - agentic-ai
  - enterprise-ai
  - enterprise-adoption
  - context-engineering
  - workflow-migration
  - knowledge-management
  - access-control
related:
  - "[[05 People/Aaron Levie]]"
  - "[[03 Research/Enterprise Adoption/Enterprise Adoption Overview]]"
  - "[[03 Research/Agentic AI/Agentic AI Overview]]"
  - "[[06 Theses/The Workflow Migration Thesis]]"
  - "[[06 Theses/The Control Surface Thesis]]"
  - "[[09 Implications/Treat Memory and Identity as Infrastructure]]"
  - "[[09 Implications/Pair Governed Runtimes With Workflow Owners]]"
---

# Aaron Levie - Enterprise Agents Context Problem

**Author:** Aaron Levie ([@levie](https://x.com/levie)), CEO of Box  
**Source:** https://x.com/levie/status/2061247380897579500  
**Posted:** Jun 1, 2026, 3:44 AM  
**Captured:** 2026-06-01 via controlled OpenClaw browser visible-text extraction  
**Quoted context:** Levie was quote-posting Tom Blomfield on the missing piece being extraction of domain knowledge from people's heads.

## Post text (verbatim)

> This is effectively the #1 problem for AI agents in the enterprise. As we go from agentic coding (where a large amount of context is in the code base, and users are technical enough to get the rest to the agent easily) to a world of knowledge work agents, the context problem becomes much more acute. We see this every day with customers at Box. For existing digital knowledge, it’s often fragmented across legacy systems or environments that don’t play nice with agents, and have access controls that don’t map to the real work that needs to be done, which become a huge hurdle for getting agents the context they need. This has to all get moved to modern, secure cloud environments. But also, companies often haven’t captured and digitized some of the critical context that agents need to work with. Decisions, processes, and workflows often live in people’s heads and tribal knowledge that need to get turned into unstructured data for agents. This is actually one of the biggest points of leverage for applied AI companies, because they can work to specialize in getting agents exactly the information and domain expertise they need. But it’s also one of the reasons why FDEs and new system integrator plays will also work so well right now. The companies that figure this out will be able to get the most out of AI going forward.

## Key points

- Levie frames enterprise-agent context as the top practical bottleneck for knowledge-work agents.
- Agentic coding has an easier context substrate because codebases already contain much of the relevant state and technical users can provide missing context.
- Knowledge-work agents face more acute context gaps because enterprise knowledge is fragmented across legacy systems, incompatible environments, and permission models that do not match the actual work.
- A second gap is uncaptured institutional knowledge: decisions, processes, workflows, and domain expertise that live in people's heads.
- Levie argues this creates leverage for applied AI companies, FDEs, and system integrators that can capture, digitize, permission, and deliver the right domain context to agents.

## Why this is in the vault

This is a clean bridge between the KB's context-discipline, memory-as-infrastructure, and workflow-migration themes.

The KB read: enterprise agents fail less from lack of raw model capability than from missing, stale, mispermissioned, or uncaptured operating context. Production adoption therefore requires modernizing knowledge stores, mapping access controls to work, converting tribal knowledge into usable unstructured data, and assigning deployment owners who can continuously shape context around workflows.

This also pairs naturally with Levie's May 28 implementation-gap signal: if agents need live enterprise context, then FDEs and integrators are not a services wrapper around AI. They are part of the context-capture and workflow-migration machinery.

## Source checks

- Original X post: https://x.com/levie/status/2061247380897579500
- Captured from logged-in controlled OpenClaw browser, June 1, 2026
- xurl API read was attempted first, but X API returned `CreditsDepleted`; browser visible text was used instead.
