Published on

AI's Impact on the Labor Market: A Discussion with a16z Partners

Authors
  • avatar
    Name
    Ajax
    Twitter

The Evolution of Software and its Impact on Labor

The discussion highlights a significant shift in how software is perceived and utilized. Initially, software was primarily a tool for managing information, essentially digitizing physical filing systems. This is evident in the early phases of software development, which focused on replacing physical filing cabinets with digital databases. Examples of this era include systems like Sabre (airline reservations), Quicken (personal finance), and PeopleSoft (HR management). These early systems improved efficiency by digitizing information but didn't drastically reduce the number of employees.

The second phase saw the rise of cloud-based software. This era was marked by the transition from on-premise servers to cloud platforms. Software like Salesforce (CRM), QuickBooks (accounting), NetSuite (ERP), and Zendesk (customer support) emerged, improving accessibility and scalability. However, these systems were still primarily focused on managing information.

Now, we're entering a third phase: AI-powered software. This is where software begins to perform tasks previously done by humans. AI agents are now capable of handling customer support, processing invoices, and performing compliance checks, signaling a significant shift from managing information to replacing or augmenting labor.

The Shift from Software to Labor Budgets

The labor market dwarfs the software market in size. For instance, the annual salary market for nurses in the US surpasses $600 billion, while the global software market is less than that. This illustrates the immense potential for software companies to tap into the labor budget.

AI enables software to perform tasks previously handled by humans, allowing companies to sell solutions that directly reduce labor costs, not just improve efficiency. This shift is captured in the concept of "Input Coffee, Output Code," where software engineers build products that automate tasks previously done by end-users, marking a significant departure from previous software generations.

Pricing Model Transformation

Traditional software pricing models based on per-user licenses may not be suitable for AI-powered software. Companies may need to adopt value-based pricing models, charging based on the labor cost reduction achieved. For example, instead of charging per support agent, a company might charge based on the number of support tickets resolved by AI.

This shift could disrupt existing software companies. Those that fail to adapt to the new pricing models could lose revenue, while companies that successfully adapt could see their revenue increase significantly.

The "Messy Inbox Problem" and AI Innovation

The "messy inbox problem" refers to the challenge of extracting information from unstructured data, such as emails, faxes, and phone recordings. Historically, humans have performed this task. AI is now being leveraged to solve this challenge.

Companies are using AI to extract information from unstructured data and automate workflows, potentially becoming the new AI-native systems of record. They often start by automating a specific task and then expanding to other areas. An example is Tenor, which started by automating patient referrals and now expands to other areas of healthcare administration.

Defensibility in the AI Era

While AI offers a strong initial differentiation, it's not enough to create a truly defensible business. The ability to use AI to solve the "messy inbox problem" may become commoditized over time. True defensibility comes from:

  • Owning the end-to-end workflow
  • Integrating deeply with other systems
  • Creating network effects
  • Becoming a platform
  • Embedding viral growth into the product

The principles that have always been important in software still apply in the AI era.

The Impact of AI on the Labor Market

AI will likely automate many repetitive tasks, but it will also create new jobs. The focus will shift to tasks requiring human connection and creativity. Examples include product managers, UX designers, and social media managers. The value of human-to-human interaction will likely increase, with people seeking genuine connections as AI becomes more prevalent.

Every white-collar job will likely have a copilot, with AI assisting people in their work and making them more efficient. Some jobs may be completely automated by AI agents.

Metrics for Evaluating AI Companies

The fundamental principles of evaluating a business remain unchanged. The focus is still on future profits, customer retention, gross margin, and fixed costs. However, the potential market size is expanding. AI enables software to enter new markets that were previously not viable by reducing labor costs, making software more affordable. The barrier to entry is also lower, making it easier to create and scale software companies, thus increasing competition.

Areas for Innovation

Niche areas are promising. The focus should be on areas where AI can provide significant improvements. Look for industries that are underserved by software. It's not advisable to try to automate everything, as some use cases are too complex or require too much integration. Instead, focus on areas where the technology is already capable of providing a 100x improvement.

Look for opportunities to disrupt old systems, such as in financial services and insurance. Consider building full-stack AI-native companies, which can have a completely different cost structure than existing companies and capture more value by owning the entire workflow. The "messy inbox problem" is a key area for innovation. Also, horizontal software opportunities still exist, as there's a need for AI-native versions of software for sales, marketing, product management, and other areas. However, it's crucial to understand the market structure and the potential for existing competitors to adapt.

Key Concepts Explained

  • Copilot: An AI tool that assists humans in their work, making them more efficient.
  • Autopilot: An AI tool that performs tasks autonomously, without human intervention.
  • Messy Inbox Problem: The challenge of extracting information from unstructured data.
  • AI-Native System of Record: A system that uses AI to manage data and automate workflows, potentially replacing traditional systems of record.
  • Vertical SaaS: Software designed for a specific industry.
  • Horizontal SaaS: Software designed for a broad range of industries.
  • NAICS Code: The North American Industry Classification System.
  • Deflationary Force: A force that drives down prices, such as technological innovation.
  • Full-Stack AI-Native Company: A company that builds its entire business around AI.