Background agents as the next workflow pattern

Summary

Aaron Levie argues that the largest agent workloads will come from long-running agents operating in the background or triggered by workflows, not from chat interfaces; read the original tweet. His example: Claude Managed Agents watch contract uploads to Box, review them, and create Linear tasks with the critical information. He sees this pattern scaling across client onboarding, invoice processing, M&A due diligence, data extraction, document generation, code writing, and data movement between systems — all tasks where an agent can work unattended.

Key Claims

• Background and workflow-triggered agents will consume far more tokens than chat-based interactions because they run continuously while users sleep or work on other things.
• Operational tasks suitable for background agents include data extraction, document review and generation, system-to-system data movement, code writing, and decision steps — work that does not require immediate human input.
• Background agents require safe code execution, tool use, compute sandboxing, and ability to connect across enterprise systems; the architecture makes or breaks the pattern.

Quotes

• "Right now the main paradigm that we think of agents in is chatting back and forth, but the biggest use of tokens will come from agents that are just always on running in the background doing work for us, or ones triggered from a workflow."
• "Agents will be working 24/7 in our workflows processing data, reviewing and generating documents, moving data between systems, writing code, accelerating decision making steps, and more."