Key Moments
The AI paradox: More automation, more humans, more work | Dan Shipper
Lenny's PodcastWant to know something specific about what's covered?
We've already dissected every moment. Ask and we will deliver (with timestamps).
Key Moments
New AI tools are making yesterday's human expertise cheap, but humans can leverage this "frozen competence" to create novel, valuable work, leading to more human involvement, not less.
Key Insights
Companies are shifting from personal AI agents to a single, powerful "super agent" for the entire organization due to the maintenance overhead of individual agents breaking frequently.
AI agents need human oversight; the 'maintenance cost' of tending to AI agents is currently the primary driver for centralizing them, not individual specialization.
The future of work will bifurcate into interacting with a company-wide agent (likely via Slack) and conducting most tasks within an AI-powered desktop environment like Codex or Claude Co-work.
SaaS tools are not disappearing; instead, AI agents will increase their user base by integrating with them, potentially saving SaaS companies' margins by using user-provided AI tokens.
The "AI job apocalypse" is unlikely, as AI commoditizes existing skills. Humans will be needed to create novel applications and manage AI systems, increasing demand for roles like "forward deployed engineers."
To succeed in the AI future, professionals should "ride the models" by being curious, playful, and applying new AI tools to their work, and focusing on unique human creativity that AI cannot easily replicate.
The rise of the "super agent" and the need for human oversight
Contrary to initial expectations of individual AI agents, the trend is shifting towards a single, company-wide "super agent." This is primarily because personal agents require significant human maintenance to function effectively, a cost that most users and companies find prohibitive. When agents break or need constant updates, the effort involved discourages adoption. The solution is a centralized "super agent" managed by a dedicated team or a "forward deployed engineer" who ensures its functionality for the entire organization. This reflects the reality that AI agents, at least currently, need human "gardening" and that the ideal architecture favors a collective, maintained system over fragmented individual solutions. For example, companies like Shopify and Ramp are adopting this model, indicating a broader industry shift.
The AI-powered desktop as the new work operating system
The second major prediction for the future of work is the emergence of AI-powered desktop environments, such as OpenAI's Codex and Anthropic's Claude Co-work, as a primary work surface. These environments integrate various tools, including in-app browsers, allowing users to perform tasks directly within the AI interface. This paradigm shift means that instead of AI being integrated into SaaS tools, SaaS tools will increasingly run within these AI environments. This is a profound change because it allows AI agents to access and interact with all the user's applications and data seamlessly, creating a parallel work buddy experience. For instance, a user can write a document in an in-app browser, with Codex or Co-work watching and assisting, researching, and even executing complex tasks like compiling reports from previous years. This integrated approach streamlines workflows, reduces procrastination on tedious tasks, and fundamentally changes how work is performed by bringing the AI agent directly into the user's workflow.
SaaS is not dead; it's evolving to collaborate with AI
The rise of AI agents doesn't signal the end of SaaS; instead, it's poised to invigorate it. The prediction is that SaaS tools will flourish, with their value shifting from solely human-centric interfaces to environments where both humans and AI agents can collaborate. Companies that embrace this will make their software compatible with agents, allowing users to leverage their own AI tokens or subscriptions. This move is significant because it alleviates the need for SaaS providers to build and maintain their own complex AI features, instead focusing on creating collaborative workflows. Agents' ability to automate tasks within SaaS tools will likely increase overall SaaS usage, turning these tools into essential components of an AI-augmented workforce. This scenario is supported by the observation that many companies, including Every, are increasing their SaaS spend even as AI adoption grows.
The "AI job apocalypse" is a myth; human roles are evolving
The fear of mass unemployment due to AI is largely unfounded. Instead of replacing humans, AI is making existing human competencies "cheap" by commoditizing them. This commoditization frees up humans to focus on higher-level, novel, and creative tasks that AI cannot easily replicate. For example, while AI can automate coding or writing, humans are needed to imbue these outputs with unique insights, ensure coherence, and direct the AI's efforts towards truly innovative solutions. This shift is creating new roles, such as "forward deployed engineers" or AI managers, who are responsible for overseeing and optimizing AI systems. The increasing volume of AI-generated output also means more work will be in reviewing and refining that output, necessitating human judgment and expertise.
Product managers and designers are poised for increased value
The future of work will highly value roles that focus on ideation, strategy, and creation. Product Managers (PMs) are predicted to thrive because AI tools can handle much of the more technical or repetitive aspects of their jobs, like writing initial product requirement documents or even shipping code. This allows PMs to focus on their core strengths: understanding user needs, defining product vision, and identifying market opportunities. Similarly, full-stack designers are expected to see a surge in demand. With AI assisting in coding and front-end development, designers can translate their creative visions into functional products more directly. They can bypass lengthy handoffs to engineers and produce unique, compelling user experiences that stand out from the "slop" of generic AI outputs. These roles are empowered by AI to be more impactful and efficient.
The evolving shape of work: output review and agent management
The nature of work is changing to accommodate AI's capabilities. One significant shift is the increased importance of reviewing and refining AI-generated output. As non-technical roles gain the ability to generate code, documents, and analyses, technical roles and quality assurance will be crucial for ensuring coherence and accuracy. This means technical experts will spend more time integrating and validating the work produced by others, including AI agents. Furthermore, managing AI agents is becoming a distinct and essential skill. This involves not just setting up agents but continuously monitoring, troubleshooting, and guiding them to ensure they perform optimally and align with organizational goals. This "gardening" of AI systems ensures their ongoing effectiveness and creates a new layer of human involvement in automated processes.
Embracing AI: Curiosity, playfulness, and "riding the models"
To navigate the evolving AI landscape successfully, individuals should adopt a mindset of curiosity and playfulness. The key is to "ride the models" by actively experimenting with new AI tools and applying them to one's own work and life. This means trying out new AI models as they are released, exploring their capabilities through iterative use, and looking for everyday problems that AI can help solve. Instead of fearing AI, individuals should embrace it as a tool to extend their powers and discover novel applications. The most effective way to stay relevant is to be at the forefront of AI adoption, discovering new use cases and influencing the development of AI tools. The notion that the "edge of AI" is solely in tech hubs like San Francisco is challenged, with the true frontier being where AI meets real human application and creativity.
Mentioned in This Episode
●Software & Apps
●Companies
●Books
●People Referenced
Navigating the AI-Transformed Workplace: Key Recommendations
Practical takeaways from this episode
Do This
Avoid This
Common Questions
Claude Code is an AI coding agent that Dan Shipper predicted would be significant for non-engineering tasks like fixing files or sorting hard drives. It was underrated because people initially focused on its coding capabilities, but its ability to interact with computers using natural language (English) proved highly valuable for a broader range of non-technical office work.
Topics
Mentioned in this video
An AI agent developed by OpenAI, now with a desktop app that integrates a browser, allowing it to act as an operating system for all work, including emails, documents, and research. Dan Shipper uses it as his daily driver.
An online markdown editor built by Dan Shipper, where he uses Codex to assist in writing documents and manage tasks. It's designed for seamless human-AI collaboration.
Every's internal email agent that works with Codex to manage and process emails, helping Dan Shipper achieve "inbox zero".
An AI-powered code editor, noted for its strong cloud implementation and focus on programmers, and recently essentially acquired by SpaceX.
A project management and note-taking software, mentioned for its use in quarterly planning at Every, where Notion agents were used to generate strategy reports and plans.
Dan Shipper's company, an AI-forward startup that has doubled in size, where everyone, including non-technical staff, uses AI tools like Codex and Co-work for daily work. They foster an early adopter culture and provide insights into AI trends.
An AI company known for Claude Code and Co-work, recognized for its early success in applying coding agents to non-technical work and developing managed agents for the cloud.
A developer platform for B2B SaaS that provides drop-in APIs for enterprise features like SSO, SCIM, and audit logs, helping startups become enterprise-ready quickly.
An AI research and deployment company, initially thought to be behind Anthropic in coding agents but recently surpassed them, especially with the Codex desktop app.
An open-source product analytics company, mentioned as an example of a website or SaaS tool that could be used inside an AI agent's internal browser.
An aerospace manufacturer and space transport services company, mentioned as having essentially acquired Cursor, indicating a growing trend of model companies integrating broader AI development tools.
A web-based platform for version control and collaborative software development. It's facing infrastructure challenges due to the exponential increase in usage by AI agents and people's agencies.
A company that automates compliance and risk management across 35+ security and privacy frameworks (like SOC 2, ISO 27001, HIPAA), helping companies earn and prove trust with customers.
Every's writing app, run by Marcus, a PM who leverages AI coding models to ship features quickly, combining his product sense with light technical skills.
A cloud-based software company, used as an example of a large enterprise where employees might face limitations in using the latest AI models due to internal policies or development cycles.
A book highly recommended by Dan Shipper, especially the last chapter, for its insights into writing, technology, and their relationship to time and the future. Required reading for employees at Every.
A multi-volume history and memoir written by Winston Churchill, recommended for its unique combination of historical account and personal experience during World War II.
A book Dan Shipper recently finished, exploring a history of ideas connecting Heisenberg, Borges, and Kant, with interesting overlaps to AI and quantum physics. Highly recommended.
A biography of Robert Moses by Robert Caro, mentioned as a current obsession for its surprisingly compelling history of New York, despite its length.
A German theoretical physicist and one of the key pioneers of quantum mechanics, known for his uncertainty principle, mentioned in relation to the book The Rigor of Angels.
A famous rock climber, known for his free solo ascents, mentioned as a comparison to Dean Potter.
An Argentinian fiction writer whose short stories are increasingly relevant to AI-related concepts, discussed in the context of the book The Rigor of Angels.
A German philosopher who is considered a central figure in modern philosophy, mentioned in connection with the book The Rigor of Angels for his intellectual contributions relevant to AI topics.
A popular figure in the "AI meditation overlap discourse" whom Dan Shipper admires for his insights, particularly a motto about dealing with hard things from a position of spaciousness and strength.
An extreme athlete known for free soloing and wingsuit base jumping, subject of the documentary The Dark Wizard, who had an extreme personality.
An American ultramarathon runner, ultra-distance cyclist, triathlete, public speaker, and author. Mentioned for his "just go for it" approach to dealing with hard things.
A previous podcast guest who suggested that people need to find their "moment of joy with AI" to truly engage with it and continue building solutions.
More from Lenny's Podcast
View all 33 summaries
100 minWhy we’re at the beginning of the AI hardware boom | Caitlin Kalinowski (ex–OpenAI, Meta, Apple)
100 minHow Anthropic, Costco, and Patagonia all build incorruptible companies | Eric Ries
88 minAI era skills: Why cultivating agency matters more than job titles | Max Schoening (Notion)
71 minSnapchat CEO: Why distribution has become the most important moat | Evan Spiegel
Ask anything from this episode.
Save it, chat with it, and connect it to Claude or ChatGPT. Get cited answers from the actual content — and build your own knowledge base of every podcast and video you care about.
Get Started Free