AI Reshapes Business: Engineer as Agent Manager, Startup Boom

Original Title: “Engineers are becoming sorcerers” | The future of software development with OpenAI’s Sherwin Wu

Lenny's Podcast: Product | Career | Growth · February 12, 2026 · Listen to Original Episode →

The coming wave of AI is not just about building better software; it's about fundamentally reshaping how businesses operate, creating unprecedented opportunities for those who understand its second and third-order effects. This conversation with Sherwin Wu, Head of Engineering for OpenAI's API and Developer Platform, reveals how AI is transforming the engineer's role from code creator to agent manager, and how this shift will catalyze a boom in specialized B2B SaaS and one-person billion-dollar startups. The hidden consequence? A widening productivity gap between AI power users and everyone else, and a critical 12-24 month window for engineers to establish new standards before the landscape solidifies. This analysis is for builders, product leaders, and strategists who want to harness AI's transformative power, offering a glimpse into a future where individual leverage is amplified and the very definition of a "startup" is redefined.

The Engineer as Sorcerer: Managing Fleets of AI Agents

The role of the software engineer is undergoing a seismic shift, moving from a hands-on coder to a conductor of AI agents. Sherwin Wu highlights that at OpenAI, nearly all engineers now use AI tools like Codex daily, with close to 100% of pull requests reviewed by AI. This isn't just about speed; it's about a fundamental change in workflow. Engineers are increasingly managing "fleets" of AI agents, akin to sorcerers casting spells, guiding and steering these AI entities to perform complex tasks. This evolution is creating a significant productivity gap, with those who master these tools opening up to 70% more pull requests.

"Engineers are becoming tech leads they're managing fleets and fleets of agents it literally feels like we're wizards casting all these spells and these spells are kind of like going out and doing things for you."

This transition is not without its challenges, reminiscent of the "Sorcerer's Apprentice" narrative, where powerful tools can go awry if not managed with skill and foresight. The next 12-24 months represent a rare opportunity for engineers to define best practices and standards in this new paradigm, before the role solidifies. The core skill is shifting from writing code to effectively prompting, guiding, and integrating AI outputs, requiring a deeper understanding of system design and agent orchestration.

The One-Person Billion-Dollar Startup: A Cascade of New Businesses

The amplified leverage AI provides to individuals is leading to the concept of the "one-person billion-dollar startup." While this headline-grabbing idea is compelling, its true impact lies in the cascading effects it triggers. Sherwin Wu posits that the ease with which a single individual can now build and scale a business will likely fuel a massive boom in startups, particularly in the B2B SaaS space. Instead of a single dominant player, we may see a proliferation of highly specialized, vertical-focused startups catering to niche needs.

"if it's easy for a person to create a one person billion dollar startup it also means it's way easier for people to just create startups in general like i actually think this will like one second order effect of this is i think there's just going to be a huge like startup boom and like small like smb style boom."

This scenario suggests a future where a few large platforms might support a vast ecosystem of smaller, agile companies. These smaller businesses, potentially reaching $10 million or $50 million in value, offer a high quality of life for their founders and employees, even if they don't fit the traditional venture-scale mold. This democratization of entrepreneurship, enabled by AI, could redefine the startup landscape, shifting focus from hyper-growth to sustainable, individual-driven ventures.

The Shifting Sands of AI Development: Beyond Scaffolding

The rapid evolution of AI models means that what works today might be obsolete tomorrow. Sherwin Wu's observation that "models will eat your scaffolding for breakfast" underscores a critical lesson for builders: focus on where the models are going, not where they are today. Early AI applications often relied on extensive "scaffolding"--frameworks, vector stores, and complex logic--to compensate for model limitations. As models become more capable, much of this scaffolding becomes redundant.

"The models will eat your scaffolding for breakfast."

This dynamic implies that customer feedback, while valuable, must be balanced with a forward-looking perspective on model capabilities. Blindly optimizing for current customer requests, such as a better vector store, might lead companies down a path that is quickly outpaced by model advancements. Instead, building with future AI capabilities in mind, anticipating improvements in areas like multi-hour task coherence and multimodality (especially audio), will lead to more durable and impactful products. This requires a strategic bet on the trajectory of AI development, understanding that the "bitter lesson" of AI is often that simpler, more scalable approaches, driven by raw model power, eventually prevail.

Actionable Takeaways for Navigating the AI Revolution

Embrace Agent Management: For engineers, shift focus from manual coding to effectively managing and prompting AI agents.
- Immediate Action: Experiment with AI coding assistants like Codex or Cursor daily.
- Longer-Term Investment: Develop skills in prompt engineering and agent orchestration.
Build for Future Capabilities: When developing AI products, design based on anticipated model advancements, not current limitations.
- Immediate Action: Identify a core product idea that leverages 80% of today's AI capabilities, with clear paths for future enhancement.
- Longer-Term Investment: Stay abreast of model advancements (e.g., multi-hour tasks, multimodality) and adapt your product strategy accordingly.
Foster Bottom-Up AI Adoption: For organizations, successful AI integration requires both top-down executive buy-in and enthusiastic bottom-up adoption.
- Immediate Action: Identify and empower internal AI enthusiasts to form an "AI evangelist" team.
- Longer-Term Investment: Implement knowledge-sharing sessions and internal hackathons focused on AI tools.
Explore Business Process Automation: Recognize the vast, often overlooked, opportunities for AI in automating repetitive business processes outside of traditional software engineering.
- Immediate Action: Identify one repetitive, manual business process within your organization ripe for automation.
- Longer-Term Investment: Invest in platforms and tools that enable high-determinism, integrated AI solutions for business operations.
Don't Fear Competition from AI Labs: Focus on building products that deeply resonate with customers; the market is large enough to accommodate many successful players.
- Immediate Action: Prioritize customer delight and product-market fit above all else.
- Longer-Term Investment: Continuously iterate on your product based on genuine user needs and feedback.
Leverage AI for Managerial Support: AI tools can augment managerial capabilities, enabling larger team management and proactive problem-solving.
- Immediate Action: Use AI tools to assist with tasks like performance review research and organizational knowledge synthesis.
- Longer-Term Investment: Explore how AI can help predict potential blockers and proactively unblock team members.
Invest in Audio and Multimodal AI: Prepare for significant advancements in audio processing and multimodal AI, which will unlock new applications beyond text and code.
- Immediate Action: Consider how audio or other modalities could enhance your current product or service.
- Longer-Term Investment: Stay informed about developments in speech-to-speech and other multimodal AI technologies.