How quickly will AI change the job market for coders?

The transition is expected to be rapid, with job roles potentially changing significantly within the next 5-10 years. While AI is already impacting how code is written, the full scope of job transformation is anticipated to unfold over this period.

What is Jevons Paradox and how does it apply to AI and software demand?

Jevons Paradox suggests that increasing efficiency in resource use leads to increased overall consumption. Applied to AI, as software becomes cheaper and easier to create, the demand for software is expected to rise significantly, though AI itself may fulfill much of this demand.

Are there any uniquely human skills that AI cannot replicate in coding?

While AI excels at generating code, uniquely human traits like deep user empathy, obsession with product excellence, and the ability to identify and frame complex problems may remain critical. These 'agency' related skills are crucial for guiding AI towards meaningful innovation.

How will AI impact knowledge work beyond software engineering, like in law or medicine?

AI is poised to decrease the cost and increase the accessibility of knowledge work in fields like law, finance, and medicine. While some roles may change, the integration of AI is becoming a competitive necessity, similar to how computers and email became standard tools.

What advice is given for founders or individuals looking to thrive in an AI-driven future?

It's crucial to stay updated with the latest AI tools and technologies, as early adopters gain a significant advantage. Additionally, honing the skill of identifying and understanding human problems to solve will be paramount, as building solutions becomes increasingly accessible.

What are the potential societal challenges of rapid AI-driven job displacement?

A major concern is the transition period, where hundreds of millions could be displaced from their jobs. Retraining and adapting to new roles is expected to be extremely painful, potentially leading to significant societal turmoil for a decade or more.

Key Moments

How AI Coding Agents Will Change Your Job

Y Combinator

Science & Technology5 min read20 min video

May 9, 2025|115,093 views|2,307|186

YC Y Combinator

Save to Pod

Key Moments

TL;DR

AI coding agents, likened to a 'combine harvester' for software, can now generate tens of thousands of lines of code, diminishing the need for traditional software engineers and democratizing the ability to build complex applications, but this rapid automation could lead to significant societal upheaval during the transition.

Key Insights

A YC founder rebuilt their 15-year-old blog and migrated content in 90 minutes using AI coding tools like Claude Code.

One third to one half of YC companies now primarily write code using AI agents, a rapid increase from approximately 0% just two batches ago.

AI coding tools enabled Tom Blomfield to build a 35,000-line application, recipes.ai, without writing a single line of code himself.

The cost of knowledge work, including software engineering, is projected to decrease drastically due to AI automation, potentially leading to widespread job displacement.

While AI can rapidly build software, the ability to identify human problems and obsess over user experience may remain a uniquely human advantage for founders.

In the next 5 years, it's argued to be the best time in history to build a company due to AI cracking open many ideas and transforming industries like law, education, and medicine.

AI as a 'combine harvester' for software engineering

The conversation frames AI coding agents as a technological revolution akin to the combine harvester in agriculture. Historically, software engineering jobs have evolved significantly, from punch cards to higher-level programming languages. The current wave of AI tools represents another major abstraction layer, empowering individuals with 'superpowers.' Tom Blomfield demonstrated this by rebuilding his personal blog, including migrating 15 years of posts, in just 90 minutes on a train using AI tools. He further built a complex, 35,000-line application with an interactive voice agent called recipes.ai, without writing any code himself, highlighting the dramatic increase in individual productivity. This revolution suggests that what once required armies of engineers may soon be achievable by small, highly capable teams or even individuals.

Rapid adoption and growing capabilities of AI coding tools

The adoption of AI coding agents in startups is accelerating rapidly. According to Y Combinator data, the percentage of companies primarily using these tools has jumped from approximately 0% two batches ago to between one-third and one-half in the current batch. While acknowledging that AI is not yet perfect, the discussion emphasizes a clear trajectory of improvement over the next 3, 6, and 12 months. This rapid progress leads to the conclusion that the argument for AI *never* being good enough to handle professional code bases is a losing one, especially when considering the exponential growth rates observed.

The 'Jevons paradox' and the future of software demand

A counterargument to the idea of job displacement is the Jevons paradox, which suggests that as the cost of a resource (like software creation) decreases, demand for it increases, potentially leading to overall market growth rather than contraction. While agreeing with this principle, the speakers argue that AI's ability to meet this surge in demand will far outpace human capacity. The analogy to electricity highlights how its cheapness led to vastly expanded use. In software, this could mean an explosion of applications catering to every conceivable need. However, the core contention remains: even with massively increased demand, the sheer productivity gains from AI mean that fewer human software engineers, as we know them today, will be required to fulfill it. The demand for software might not be infinite, but it is very large, potentially leading to a future of on-demand, custom-built, ephemeral software solutions.

Disruption beyond software engineering

The impact of AI extends beyond software engineering to other knowledge work professions like law, medicine, and accounting. Founders are already building AI tools tailored to these domains, and adoption within these industries is beginning to overcome initial skepticism. Historically, professions like law operated on billable hours, creating a disincentive for efficiency-boosting tools. However, the pervasive worry across industries about AI's impact is driving adoption. Just as not using a computer or email is now unthinkable, AI proficiency is likely to become a competitive necessity, making it a disadvantage for professionals who do not embrace these tools. While widespread adoption may lag slightly behind software, this shift is anticipated within the next year or two.

The potential for human agency and unique problem-solving

Despite the power of AI agents, the discussion posits that uniquely human characteristics like agency, taste, and the obsession with solving specific user problems might remain critical. The best software products are often driven by individuals or small teams with a deep, personal commitment to user experience – a level of obsession that is difficult to program into AI. In a future brimming with AI-generated solutions, the ability to identify genuine human problems and empathize with users could become the most valuable skill. This 'human element' is seen as a potential differentiator and a key area where individuals can retain high agency, especially for founders looking to build impactful products.

An unprecedented era for founders and startups

The current technological landscape, powered by AI, is presented as the most opportune time in history for aspiring founders. AI tools democratize creation, allowing individuals and small teams to achieve what previously required large organizations and significant capital. Founders can now build complex products faster, reach scale more quickly, and potentially achieve profitability with less external funding. The transformation of industries like law and medicine, which were previously less receptive to software solutions, opens up vast new markets and opportunities. This confluence of readily available building tools and emerging market demand makes starting a company today a uniquely exciting prospect.

Navigating the transition: challenges and recommendations

The rapid pace of AI-driven change presents a significant challenge: the transition period. While a future of abundance and knowledge democratization is envisioned, the potential for hundreds of millions of people to be displaced from their current roles could lead to severe societal turmoil. The retraining process is expected to be painful and lengthy. Protective measures, similar to regulatory hurdles faced by self-driving cars despite their proven safety, may emerge from professional bodies seeking to safeguard existing jobs. Despite these anxieties, the consensus is that moving towards this AI-integrated future will ultimately lead to a better human experience, characterized by abundance and expanded capabilities. For individuals, staying updated with AI tools and honing problem-identification skills are paramount to thriving in this evolving landscape. The ability to understand and connect with human problems is highlighted as a skill that will become even more important as building software becomes easier.

Mentioned in This Episode

●Products

●Software & Apps

●Companies

●Organizations

●Concepts

●People Referenced

Rate of AI Adoption in YC Founder Codebases

Data extracted from this episode

Timeframe	Percentage of Companies Using AI for Most Code
Last 6 months	33%-50%
Two batches ago (approx. 1 year ago)	25%
Before that	0%

Common Questions

The consensus is that while AI will dramatically change software engineering, completely replacing human engineers is unlikely. New roles will emerge for those who can effectively manage and direct AI coding agents, focusing on higher-level problem-solving and overseeing AI-generated code.

Topics

AI & Machine Learning Technology & Innovation Future Of Work Knowledge Work Software Engineering AI Coding Agents Startup Founders Career & Skills Tech Industry Trends

Mentioned in this video

Software & Apps

Replit

A collaborative coding environment and learning platform used by Tom Blafield to experiment with building simple games, highlighting the ease of use of modern development tools enhanced by AI.

tomblonfield.com

Tom Blafield's personal blog, which he rebuilt using AI coding tools, serving as a case study for the capabilities of these new technologies.

recipes.ai

A project involving an interactive voice agent that Tom Blafield built using AI coding tools. It evolved into a substantial project with 35,000 lines of code and thousands of users.

ChatGPT

Referenced as a potential interface for generating on-demand custom software solutions for specific user problems, highlighting a future where ephemeral programs are created and utilized as needed.

Windsurf

An AI coding tool mentioned by Tom Blafield as part of his suite of tools used to rebuild his blog and develop new applications, indicating the growing ecosystem of AI-assisted development platforms.

Lovable

A no-code tool mentioned as one of the platforms Tom Blafield used to build simple games, demonstrating the accessibility of AI-powered development tools.

Cursor

An AI-first code editor that assists developers with coding tasks, mentioned by Tom Blafield as a tool he used during his experimentation with AI coding agents.

Concepts

Jevons Paradox

Mentioned as a counterargument where falling costs of a resource (like software development via AI) could lead to increased overall demand, though the discussion posits AI, not humans, would meet this demand.

People

Tom Blafield

Co-host of The Breakdown podcast, who shared his personal experience using AI coding agents and his provocative tweets about the future of software engineering.

Products

Combine Harvester

Used as an analogy to describe a technological revolution that drastically increased food production while significantly reducing the need for human farmers, paralleling the potential impact of AI on software engineering.

Companies

Lora

A Swedish company that went through Y Combinator, mentioned as an example of a startup successfully building tools for the legal industry, contrary to the received wisdom that lawyers don't buy software.

Y Combinator

A startup accelerator mentioned for its cohort of founders who are early adopters and developers of AI tools, indicating the frontier of AI development and adoption within the startup ecosystem.