Statement
Software for Agents
- The next trillion internet users are projected to be AI agents rather than people, creating an urgent need to develop software that meets agent requirements immediately.
- Future agents are expected to perform tasks such as web browsing, research, purchasing, and managing legacy CRM systems, necessitating a shift from human-centric visual interfaces to machine-readable protocols like APIs, MCPs, and CLIs.
- Software categories currently designed for humans are characterized as slow, inconsistent, and brittle for agents, requiring thorough documentation to enable programmatic discovery, sign-ups, and tool usage without human intervention.
- Major software categories are anticipated to be rebuilt from the ground up specifically for agents, as this "agent-first" infrastructure will not originate from incumbents adding support to existing platforms but rather from startups treating agents as first-class citizens.
- The primary market opportunity is identified as building the foundational software that agents depend on, rather than developing agents themselves.