AI summary by TwinMind

How AI Coding Agents Will Change Your Job

Y Combinator

19m 11s

Tom's Controversial Tweet and Backlash

  • Tom compared software engineers to "highly paid organic farmers" tending crops by hand
  • Predicted "combine harvester" technological revolution coming to software engineering
  • Suggested humanity will produce more software but employ fewer engineers
  • Tweet upset software engineering community, particularly certain circles
  • Followed up with longer blog post explaining reasoning

Personal Experience with AI Coding Tools

  • Started using no-code tools (Lovable, Replit) 2 months ago to build games
  • Tools dramatically improved since 6-9 months prior attempt
  • Rebuilt personal blog (tomblomfield.com) in 90 minutes on train using Claude Code
    • Set up hosting, wrote new blogging software, migrated 15 years of blog posts
  • Built recipes.ai - 35,000 lines of code, thousands of users, full interactive voice agent
    • Wrote zero lines of code personally
    • Stopped reading code after first 5,000 lines
    • Would prompt, auto-accept, make coffee, return to completed features
  • Tools made him 10x more productive than when professionally coding 10 years ago

YC Company Adoption Data

  • Current batch: 1/3 to 1/2 of companies primarily write code using AI tools
  • Last batch: 25% used these tools for most code
  • Two batches ago: approximately 0%
  • Rapid adoption by YC founders (frontier users with smaller, less legacy codebases)

Responses to Automation Concerns

Response 1: "AI Not Good Enough"

  • Dismissive of current capabilities for professional codebases
  • Counter: Rate of improvement suggests tools will become capable enough
  • Historical pattern: innovations start as "toys" then rapidly improve (Clayton Christensen's Innovator's Dilemma)

Response 2: Jevons Paradox Argument

  • As software cost decreases, demand increases exponentially
  • Similar to electricity - cheaper cost leads to higher total usage/spending
  • Counter: AI will fulfill increased demand, not humans
  • Combine harvester analogy: 10x food productivity increase, 1000x reduction in farmers per unit
  • Software automation may be more pronounced than agriculture

Future of Software Engineering

  • Current software engineering jobs won't exist in 5-10 years
  • Demand for people who can "wrangle AI coding machines"
  • Job will be dramatically different from today
  • Historical precedent: early programmers (punch cards, machine code) vs. modern software engineers
  • AI represents another abstraction layer allowing humans to operate at higher level

Human Monopoly Areas

  • Agency: figuring out what problems to solve
  • Taste: determining quality and user experience
  • Best software typically has single obsessed human behind the team
  • Challenge: programming AI to have obsession with solving specific problems
  • Current methodologies don't yet exist for AI obsession

Opportunities for Founders

  • Best time in history to start a company
  • High-agency individuals get "superpowers" from AI tools
  • Faster path to $1M, $10M, $100M revenue
  • Less capital required, faster profitability
  • May only need seed round, skip Series A/B
  • Excellent current opportunity, uncertain 5-10 year forecast

Impact on Knowledge Work

Legal Industry (Lora Example)

  • Swedish YC company from 18-24 months ago
  • Traditional belief: lawyers never buy software, bill by hour so efficiency hurts revenue
  • Every industry now worried about AI impact
  • Investors/boards demanding AI strategies
  • Becoming competitive disadvantage to not use AI tools
  • Similar to not using computers/email

Other Industries

  • Finance, medicine facing similar pressures
  • Not yet competitive disadvantage but approaching that point
  • Cost of knowledge work decreasing dramatically
  • Huge consumer surplus benefit

Protected Roles and Barriers

  • Physical work protected: surgeons, brick layers, plumbers, electricians
  • Industries may create regulatory barriers
  • Example: requiring doctors to prescribe drugs even when AI is better
  • Self-driving cars precedent: provably safer but regulatory challenges
  • Trade bodies acting as protectionist gatekeepers

Transition Period Concerns

  • Hundreds of millions potentially displaced
  • Retraining extremely painful
  • Societal impact and turmoil for 10-20 years during transition
  • Speed of change unprecedented
  • Long-term: future of abundance, cured diseases, new sources of purpose

Advice for Current/Future Founders

Stay Current with Tools

  • Keep up with latest AI tools even if not perfect for your industry yet
  • Betting they will cross tipping point and become extremely useful
  • Early adopters get several-year advantage
  • Enables earning potential and career growth

Focus on Human Problems

  • Get good at identifying problems to solve
  • All B2B SaaS ultimately serves human experience
  • Understand people, talk to them about real problems
  • This skill becomes more important when building becomes easier

Future Product Quality

  • Smaller teams (2-4 engineers vs. 40) can build equivalent products
  • Design quality will improve dramatically
  • Bad experiences often result from unclear ownership between teams/people
  • Single person/small group ownership creates higher quality experiences
  • Excited to use these better products

Industries Being Transformed

  • Law, education, medicine historically not big software buyers
  • Next 5 years will see major transformation
  • Many new opportunities created by AI capabilities
  • No more exciting time to build a company than right now

TwinMind

Get summaries like this for your own meetings

TwinMind takes notes in the background during meetings, lectures, and calls, then writes the summary for you.

Free plan to get started

No bot joins your calls

Phone, desktop & Chrome apps

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome