The Real Economy Hasnt Felt Ai Yet Aaron Levie

More capable AI can remain unused when companies cannot connect it to the information and processes required for work. That is the argument Aaron Levie of Box makes in an interview article published by Browserbase, a company that sells browser infrastructure for AI agents. He describes a widening gap between what models can do and what enterprises have put into production.

The article points to fragmented data, access controls that limit what an agent can reach, and slow procurement and security reviews as obstacles that model improvements do not automatically remove. On this account, putting AI to work requires integration and changes to existing workflows as well as capable models. The argument explains an adoption bottleneck, rather than measuring the economy-wide effects of AI, and Levie's estimates are his own views, offered in a vendor's write-up.