How AI Coding Agents Will Change Your Job?

om Blomfield and David Lieb, both Group Partners at Y Combinator (YC), explored how AI coding tools are revolutionizing software development. Blomfield, co-founder of Monzo, a UK challenger bank that raised over £500 million and serves 10% of the UK population, brings his experience scaling a fintech giant.

Lieb, co-founder and CEO of Bump (YC S09), a mobile app used by over 150 million people for photo sharing, offers insights from building consumer tech.

Their conversation, sparked by Blomfield’s controversial tweets, highlights how AI empowers small, high-agency teams to achieve what once required armies of engineers, with far-reaching implications for startups and knowledge work.

The Tweet That Ignited a Firestorm

Blomfield’s recent tweet comparing software engineers to organic farmers, with AI coding tools as the “combine harvester,” stirred significant debate.

He suggested that, like the harvester that boosted food production while reducing the need for farmers, AI will exponentially increase software output but diminish traditional engineering roles.

Blomfield’s AI-Powered Experiments

Blomfield’s perspective is grounded in hands-on experimentation. Eager to reconnect with product-building, he started with no-code platforms like Lovable and Replit, creating simple games.

Impressed by their advancements, he progressed to tools like Cursor, Windsurf, and Claude Code.

In a remarkable feat, he rebuilt his 20-year-old Tumblr blog, tomblomfield.com, in 90 minutes on a train, handling hosting, new blogging software, and migrating 15 years of posts.

His most ambitious project, recipeninja.ai, demonstrates AI’s potential. This 35,000-line codebase, supporting thousands of users and an interactive voice agent, was built without Blomfield writing a single line of code.

After the first 5,000 lines, he stopped reviewing the code, relying on prompts and auto-accepting changes.

As a former professional developer, he was “astonished” at how these tools made him “10 times more powerful” than a decade ago, highlighting their transformative impact.

AI Adoption Among YC Startups

At YC, Blomfield and Lieb have witnessed a surge in AI coding tool adoption. In their latest batch, 33-50% of companies primarily use these tools, up from 25% in the prior batch and nearly 0% two batches ago.

This rapid shift, particularly among founders with smaller, less legacy-heavy codebases, signals the tools’ growing reliability. Lieb noted, “The good founders are doing it,” indicating that AI coding is becoming a hallmark of cutting-edge startups.

Can AI Really Replace Engineers?

Critics of Blomfield’s tweet offered two main objections. First, some argued that AI tools are inadequate for professional-grade codebases, suitable only for “toy apps.” Blomfield and Lieb counter that AI’s rapid improvement—driven by better tool calling, interfaces, and models—makes this skepticism shaky.

Drawing on Clayton Christensen’s Innovator’s Dilemma, they argue that AI tools, like early disruptive products, start small but quickly surpass incumbents.

The second critique cites Jevons Paradox: as software becomes cheaper to produce, demand will soar, potentially preserving jobs. Blomfield agrees that demand could increase 10x or 100x but contends that AI, not humans, will fulfill it.

He envisions a future of on-demand, ephemeral software tailored to individual needs, with AI handling most development, reducing the need for traditional engineers.

The Future of Software Engineering

Blomfield predicts that traditional software engineering roles may disappear within 5-10 years. While demand for “smart people who wrangle AI coding machines” will remain, the job will be fundamentally different.

Lieb compares this to the evolution from punch-card programmers to modern engineers, suggesting AI is another abstraction layer, elevating humans to higher-level roles.

They speculate that human strengths like agency (identifying problems) and taste (ensuring quality) may persist, as AI struggles to replicate the obsessive drive of visionary founders.

Impact on Knowledge Work

AI’s influence extends beyond software. Startups like Lora, a YC alum, show that even lawyers, once resistant to software, are adopting AI to stay competitive. Investors now demand AI strategies, making adoption essential across law, finance, and medicine.

While physical trades like surgery may be less affected, regulatory barriers could delay AI’s impact in fields like medicine, even when AI outperforms humans. This shift will make knowledge work cheaper and more accessible, creating a “consumer surplus” but requiring professionals to adapt.

The Transition and a Future of Abundance

Lieb is optimistic, noting that technological progress has consistently improved human life. However, Blomfield warns of a turbulent transition, with millions potentially displaced in a 10-20-year period.

Retraining may be challenging, and industries may erect barriers to protect jobs. Yet, both envision a future where AI solves major problems, like curing diseases, and humans find new sources of purpose.

Advice for Future Founders

Blomfield and Lieb offer detailed guidance for current and aspiring founders navigating this AI-driven landscape:

Master AI Tools Early: Blomfield urges founders to stay current with tools like Cursor and Claude Code, even if they’re not yet perfect for their industry. “I’m betting a lot of money that at some point they will cross that tipping point,” he says.

Early adopters will gain a multi-year advantage, enabling them to “earn a lot of money” and build strong careers. Staying ahead of the curve is critical as these tools rapidly evolve.

Prioritize Human-Centered Problem Solving: Lieb emphasizes that as building software becomes easier, the ability to identify and solve human problems will be the defining skill for founders.

Embrace Small, High-Agency Teams: AI tools enable teams of 2-4 to accomplish what once required 40 engineers. Lieb is excited about the “second-order effect”: smaller teams mean clearer ownership, reducing the poor design and bad experiences often caused by unclear responsibilities in large teams.

“One person or a very small group can really be the owners of a high-quality experience,” he says, predicting a surge in well-designed, user-friendly products.

Capitalize on Unprecedented Opportunities: Blomfield believes “the next 5 years is the best time in the history of humanity to build something from scratch.” AI is unlocking opportunities in industries like law, education, and medicine, which were previously resistant to software.

Founders can now achieve profitability with less capital, potentially bypassing traditional funding rounds like Series A or B. This democratization of building makes it an “exciting time” to launch a company.

Tom Blomfield and David Lieb’s conversation illuminates a transformative moment in software development. AI coding tools are empowering small, high-agency teams to achieve extraordinary results, reshaping startups and knowledge work.

While the transition may be challenging, the potential for innovation and abundance is unparalleled. For founders, the path forward involves mastering AI tools, focusing on human problems, leveraging small teams, and seizing the unique opportunities of this era.

‍