Summary
Highlights
Dan Shipper, CEO of Every, is interviewed about his remarkably accurate past predictions regarding Claude Code for non-engineering work. His company, Every, has doubled in size in the past year, attributing this growth to their AI-forward approach, with all employees, including non-technical staff, utilizing tools like Codex and Claude Code. The discussion centers on how work will change, the evolving nature of job roles, and who will thrive in this AI-driven future.
Shipper predicts a bifurcation in how people use AI agents. Firstly, every company will have at least one 'super agent' that employees can delegate tasks to, often integrated with communication platforms like Slack. Secondly, much of the work will occur within environments like Codex or Cloud Co-work, which will become the 'operating systems' for many tasks, effectively turning them into platforms where SaaS tools operate.
A crucial insight is that AI agents, especially personal ones, require human maintenance and care. This leads to the concept of 'super agents' managed by a forward-deployed engineer responsible for ensuring the agent’s functionality for the entire company. This new role highlights that automation doesn't eliminate human work but shifts its nature, creating new engineering challenges in building robust and user-friendly AI systems.
Contrary to popular belief, Shipper is bullish on SaaS. He argues that AI agents will increase the number of SaaS users rather than replace them. SaaS companies will need to adapt their products to be usable by both humans and agents, focusing on features like advanced CLIs, clear HTML structures, and seamless integration within AI environments like Codex. This shift could lead to cost savings for SaaS providers as users bring their own AI tokens.
The trend is moving away from purely CLI-based interactions towards more intuitive graphical user interfaces (GUIs) where humans and AI agents collaborate. This necessitates a new kind of software design that provides visibility into both human and agent actions, approval workflows, and robust logging. This parallel collaboration is seen as incredibly powerful, leading to innovations like agent-generated bug reports that are more detailed and actionable than human-created ones.
AI will enable non-technical employees to perform technical tasks, leading to a surge in activities like pull requests. This increases the demand for technical people to manage and integrate this output coherently. Shipper also predicts a shift where AI-generated content, particularly for internal documents and emails, will become commonplace and accepted, as long as humans maintain oversight and accountability for the content.
Shipper is particularly bullish on product managers (PMs) and full-stack designers. PMs, with their strong product sense and understanding of user needs, can leverage AI coding models to build and ship products faster without needing extensive engineering teams. Similarly, full-stack designers can use AI as a 'super tool' to bring their creative visions to life, directly building and iterating on designs.
While AI makes 'yesterday's human competence' cheap and commoditized, it creates new opportunities for humans to focus on innovative and unique problem-solving. This means continuously asking, 'How do I use this to make something new and interesting?' The human ability to frame problems, articulate needs, and integrate AI insights into novel solutions will remain highly valuable.
To succeed, individuals must 'ride the models' by actively engaging with AI tools, experimenting with new models, and integrating them into their workflows. This involves being curious, playful, and consistently exploring how AI can extend one's capabilities. Shipper emphasizes that the 'edge of AI' is where it meets real human application, not just in its creation in tech hubs. The accessibility of advanced AI models means anyone can be at this 'edge of discovery.'
Shipper encourages listeners to experiment with tools like Codex or Cloud Co-work, find 'moments of joy' with AI, and apply AI to solve real-world problems. He recommends books like Annie Dillard's 'The Writing Life' and Churchill's 'History of World War II,' and highlights 'The Dark Wizard' documentary. His life motto, 'Do things worth writing about and write things worth reading,' underscores the importance of meaningful creation. He concludes by praising Codex as an underrated AI tool that has transformed his work.