The software industry's focus on fully autonomous AI systems may be misplaced, according to Robert Englander writing on O'Reilly Radar. Rather than pursuing agents that operate independently to replace workers or write code, Englander argues that conversational interfaces represent a more promising path forward for software development.
Current industry momentum emphasizes autonomous agents capable of performing tasks without human intervention. Startups have emerged specifically to build these kinds of systems, reflecting widespread belief that independent AI operation represents the future of software. These agents promise to handle everything from worker replacement to code generation to application management on behalf of users.
Englander challenges this direction. His position centers on conversational software as an alternative model that could better serve users and businesses. Rather than systems operating autonomously in the background, conversational approaches keep humans in the loop through dialogue-based interfaces. This framework allows users to maintain control while leveraging AI capabilities.
The distinction matters operationally. Autonomous systems require extensive specification of goals and constraints upfront, then execute independently. Conversational systems instead engage in ongoing dialogue, allowing users to guide decisions, ask questions, and adjust direction in real time. This collaborative model addresses concerns about accountability and control that autonomous approaches create.
Englander's argument arrives as the industry experiences a boom in agent-focused development and investment. His perspective suggests a fundamental strategic question: whether AI software innovation should prioritize systems that operate independently or those that enhance human decision-making through dialogue.
The implications extend across software categories. Development tools, business applications, and consumer software could all benefit from conversational design rather than full autonomy. Users retain understanding of system decisions and can intervene when necessary, while still accessing AI capabilities that would be difficult or impossible to implement manually.
This framing does not reject AI capabilities outright. Instead, it proposes a different integration model where human and machine intelligence work together through natural language interaction rather than separation of