Every CEO Dan Shipper: All AI Agents Need Human Supervision, SaaS is Not Dead
Dan Shipper, CEO of Every, stated in Lenny's Podcast that he runs the most AI-native company today, doubling the number of employees within a year while operating 6 AI products. He believes that "fully autonomous AI" is merely a marketing term, and all agents require continuous human supervision, context provision, and problem-solving.
Shipper pointed out that the interface for knowledge work is collapsing into Codex and Claude Co-work. The companies and individuals that will succeed in the future are those who can continuously harness each new generation of models to accomplish tasks that were previously impossible. PMs and designers are becoming even more important in the AI era, with designers now able to submit Pull Requests directly.
He emphasized that CEOs must personally use the latest models weekly, or the company will lose its direction. SaaS spending is increasing rather than decreasing, as agents will exponentially increase the demand for SaaS tools. What truly matters is "dancing with the models," continuously pushing new models to the forefront of practical work.
Source: Public Information
ABAB AI Insight
Dan Shipper, as CEO of Every, has long practiced deep applications of AI in content and productivity tools. This interview systematically summarizes the cognitive shift after a year of intensive AI practice, moving from early beliefs in personal agents to a hybrid model of company-level central agents plus human pre-deployment.
On the capital path, Every has achieved efficient expansion through extensive hiring of human supervisors for agents, while increasing SaaS tool spending, indicating that AI has not replaced human jobs but rather amplified the demand for high-quality PMs, designers, and engineers. Funding is concentrating on companies and talents that truly understand the "human-machine symbiosis" workflow.
This view is similar to the transition from CLI to GUI in the early software era, and parallels the current shift from pure automation fantasies to the reality of human-machine collaboration. The AI application layer is currently transitioning from the narrative of "replacing humans" to the pragmatic stage of "humans + agents dancing together."
Essentially, this is a correction of technological replacement: AI commodifies yesterday's human capabilities, but the real barrier lies in humans continuously providing context, taste, and first-principle thinking, shifting pricing power from pure model capabilities to teams and organizations that can effectively harness the models.
ABAB News · Cognitive Law
All agents need human caretakers; true intelligence has never been independent, but symbiotic.
If CEOs do not personally engage with the models, the company will come to a halt; the boundary of leadership equals your distance from the latest models.
Models consume yesterday's capabilities, while humans are responsible for tomorrow's unseen tasks; only those who ride the wave of new models will not be replaced.