If we want them to do Knowledge Work, design them as Knowledge Agents — Benjamin Clavié, Mixedbread

AI agents should be designed as knowledge workers rather than mere coding assistants to handle the ambiguity and complexity of real-world information processing. While coding agents benefit from durable, structured cues, knowledge work requires navigating diffuse, context-dependent information where search intent is paramount. Effective agentic systems must move beyond simple lexical search primitives like BM25 toward multi-modal, orchestrated architectures that mirror professional organizational structures, such as the relationship between partners and paralegals. By breaking down complex, open-ended queries into manageable sub-tasks for specialized agents, systems can significantly reduce the "oracle gap" and improve performance. Ultimately, tools should not be treated as neutral add-ons but as strategic components co-designed with agents to overcome performance ceilings and scale knowledge-intensive workflows efficiently.