How AI Coding Agents Will Change Your Job

Key Concepts

• Coding Agents: AI tools capable of automating software development tasks.
• Jevons Paradox: As the cost of a resource decreases, demand for that resource increases.
• High Agency Individuals: People who can leverage AI tools to significantly amplify their capabilities.
• Abstraction: Hiding complex implementation details to simplify usage, as seen in programming languages.
• On-Demand Custom Software: Ephemeral programs generated by AI to solve specific user problems.
• Consumer Surplus: The difference between what consumers are willing to pay and what they actually pay, indicating increased value.

The Automation of Software Engineering

Tom Blomfield discusses his provocative tweet comparing software engineers to "highly paid organic farmers" and the impending "combine harvester" of AI-driven coding agents. He argues that these agents, while not perfect today, are rapidly improving and will have significant consequences for software engineering.

• The "Combine Harvester" Analogy: Just as the combine harvester dramatically increased food production while reducing the number of farmers, AI will automate much of software engineering, leading to increased software output with fewer traditional software engineers.
• Personal Experience: Blomfield recounts his experience building games and rebuilding his blog (tomblonfield.com) in 90 minutes using tools like Claude Code. He then built recipes.ai, a 35,000-line project with an interactive voice agent, without writing a single line of code himself.
• Increased Productivity: Blomfield, a former software developer, found himself "10 times more powerful" using these AI tools than he was when actively coding professionally.
• YC Portfolio Companies: He notes that a significant portion (a third to a half) of Y Combinator companies are now primarily using AI-assisted coding, a dramatic increase from previous batches.

Responses to the Automation Argument

Blomfield addresses two main counterarguments to his thesis:

1. "AI is not good enough": He dismisses this, arguing that the rapid rate of improvement in AI models and tooling makes it inevitable that AI will become capable of writing complex code. He uses the "innovator's dilemma" as an analogy, noting that disruptive technologies often start as "toys" before rapidly surpassing incumbents.
2. "Jevons Paradox will save software engineers": While agreeing that the demand for software will increase as costs decrease, he argues that AI will fulfill this increased demand, not humans. He believes the productivity gains from AI will far outweigh the increase in demand, leading to fewer traditional software engineering jobs.

The Future of Software Engineering and Knowledge Work

Blomfield envisions a future where software engineering jobs as we know them will not exist. Instead, there will be a demand for people who can "wrangle" AI coding machines.

• Abstraction and Higher-Level Agents: He argues that AI is simply another layer of abstraction, allowing humans to operate at a higher level.
• On-Demand Custom Software: He foresees a future of ephemeral, custom software generated by AI to solve individual user problems.
• Impact on Knowledge Work: He extends the argument to other knowledge work domains like law, medicine, and finance, where AI is increasingly being adopted.
• Lorra Example: He cites Lorra, a YC-backed legal tech company, as an example of a company successfully disrupting a traditionally resistant industry.
• Competitive Disadvantage: He argues that not embracing AI will soon become a competitive disadvantage in many industries.

The Role of Humans and the Transition Period

The discussion explores the unique capabilities of humans and the potential challenges of the transition to an AI-driven future.

• Human Agency and Taste: The importance of human agency in identifying problems and ensuring product quality is emphasized. The question of how to program AI to be "obsessed" with solving problems is raised.
• Potential for Societal Turmoil: The potential for mass displacement of workers and the challenges of retraining are acknowledged.
• Protectionist Measures: The possibility of professional organizations acting as gatekeepers to protect jobs is discussed.

Advice for Founders and Future Skills

Blomfield and Dave share advice for founders and individuals navigating this changing landscape:

• Stay Up-to-Date: Keep abreast of the latest AI tools and technologies.
• Focus on Human Problems: Develop skills in identifying and understanding human needs and problems.
• Smaller Teams, Better Design: AI-powered tools enable smaller teams to build high-quality products with better design.
• Best Time to Build: They conclude that now is the best time in history to build something from scratch, given the opportunities created by AI.

Conclusion

The conversation paints a picture of a rapidly changing world where AI is poised to transform software engineering and other knowledge work domains. While acknowledging the potential challenges of this transition, the speakers express optimism about the future and emphasize the importance of adapting to and embracing these new technologies. They highlight the unprecedented opportunities for founders and individuals who can leverage AI to solve human problems and create innovative solutions.