AI coding tools have been advancing at breakneck speed, bringing massive changes to the software development landscape. A provocative tweet from an investor recently sent shockwaves through the developer community, comparing software engineers to "high-end organic farmers who farm by hand" and predicting that a technological revolution like "combine harvesters" is coming soon.
Real-World Experience Shows the Power of AI Coding Tools
This isn't just speculation – there's real evidence backing up these claims. Tom Blomfield, a partner at Y Combinator, recently shared his incredible experience using AI coding tools over the past two months.
He started by experimenting with no-code tools like Lava World and Replit to build simple games. Tools that were mediocre just 6-9 months ago had become surprisingly impressive.
But the real shocker came with his next project. He completely rebuilt his personal blog that he'd been running on Tumblr for 20 years – and did it all in just 90 minutes on a train. Using Claude, he handled everything from hosting setup to writing new blogging software to migrating 15 years' worth of blog posts.
The truly mind-blowing project was RecipeNinja.ai. This is a full-fledged web application with 35,000 lines of code, used by thousands of people, complete with an interactive voice agent. Here's the kicker: he didn't write a single line of code himself. After the first 5,000 lines, he didn't even read the code. He'd just input prompts, auto-accept the changes, grab a coffee, and come back to find new features fully implemented.
Rapid Changes in the Startup Ecosystem
This transformation isn't just personal anecdotes. Data from Y Combinator's portfolio companies shows just how fast things are changing:
- 2 years ago: 0% of companies primarily using AI coding tools
- 1 year ago: 25%
- Today: 30-50%
These aren't toy apps either. We're talking about real businesses generating actual revenue using these tools as their primary development method.
Common Counterarguments and Responses
Two main counterarguments have emerged against these claims.
The first is "AI will never be good enough to write professional-grade code." But this suffers from the fallacy of judging the future based on the present. When you consider improvements in tool-calling capabilities, optimization of usage patterns, and advances in underlying models, this argument loses steam. As Clayton Christensen explained in "The Innovator's Dilemma," all disruptive products start out looking like toys but improve rapidly to overtake existing solutions.
The second counterargument is more sophisticated. It invokes "Jevons Paradox," arguing that as software development costs decrease, demand will increase even more, expanding the overall market. It's like how cheaper electricity leads to more electricity consumption.
This logic is sound, but the key question is: who will meet that increased demand? When combine harvesters were introduced, food production increased 10x, but the number of farmers decreased 1000x. We're likely to see something similar in software.
The Future Software Development Landscape
The future points toward an era of on-demand, custom software. Users will bring problems to ChatGPT, and small programs will be created instantly to solve those specific issues. Temporary programs with custom UIs will solve problems and then disappear.
In this changing landscape, the current software engineer role will likely disappear within 5-10 years. There will still be demand for smart people who know how to operate AI coding machines. We might still call them software engineers, but their work will be completely different from today.
This kind of change isn't new. Early computer programmers wrote machine code and created punch cards – completely different from what today's software engineers do. Object-oriented programming languages and various tools allowed humans to work at higher levels of abstraction. AI coding tools are just another step in this abstraction ladder.
What Remains Uniquely Human?
So what can only humans do? There are a few candidates.
First is "agency" – the ability to decide what problems to solve. Second is "taste" – the ability to judge what's good or bad.
Looking at the best software we use daily, most have someone behind the product team who's obsessed with making it perfect. In a future where AI agents roam the world looking for good ideas and problems, who will have that obsession? How do you program that kind of obsession into AI? There's no clear methodology for this yet.
A Golden Age for Entrepreneurs
In the short term, though, these tools provide superpowers to highly driven individuals. If you're thinking about starting a company, there's never been a better time. You can do so much more than you could even last year.
Software ideas are reaching $1 million, $10 million, and $100 million in revenue faster than ever. You can reach profitability with less capital, and you might not even need Series A or B funding after your seed round.
Teams of 2-3 engineers can now do what previously required 40 people. This should dramatically improve product design quality. Much of bad design and user experience comes from unclear team interfaces and ownership – problems that will disappear.
Expansion to Other Knowledge Work
This transformation isn't limited to software development. Similar changes are happening in law, medicine, accounting, and other knowledge work fields.
For example, a Swedish team created Gora, developing AI tools for the legal field. The conventional wisdom was "lawyers never buy software" because hourly-billing lawyers saw efficiency tools as potential income reducers.
But now every industry is worried about AI's impact. Every investor asks about AI's effect on business in board meetings, and every board member asks management the same question. Not using AI is becoming a competitive disadvantage.
Just as you wouldn't hire a lawyer who doesn't use computers, or someone who doesn't use email, soon professionals who don't use AI tools will be left behind.
Challenges and Opportunities in Transition
The future this change brings will certainly be better. As knowledge work costs continue to fall, everyone will have access to better services. We can imagine an abundant future where all diseases are cured and humanity finds new ways to discover purpose in life.
But the biggest concern is the pain of transition. This change could happen very quickly – that's what's different from the past. The process of hundreds of millions of people losing jobs and retraining for new careers will be extremely painful. This social shock and disruption could last 10-20 years.
Jobs requiring physical work – surgeons, bricklayers, plumbers, electricians – will be relatively safe. But at-risk industries might erect barriers. Like regulations requiring only doctors to prescribe medicine, we might see AI blocked by regulations even when it performs better. It's similar to how autonomous vehicles, despite being proven safer than human drivers, face deployment difficulties due to regulations.
How to Prepare for the Future
So how should entrepreneurs and aspiring entrepreneurs prepare?
First, keep up with the latest tools. Even if they're not perfect now, they'll eventually cross a threshold and become extremely useful. People at the forefront of these tools will have advantages that could last years, allowing them to make significant money and build great careers.
Second, developing the ability to identify human problems is crucial. Every B2B SaaS company ultimately exists to provide better experiences for individual humans. Every B2B company eventually serves companies that serve consumers. Understanding people, discovering problems, and talking to people to identify their real issues will be the most important skill for entrepreneurs in a future where building things becomes easy.
The Best Time in History to Start Something
In conclusion, this is the best time in human history to build something from scratch. AI has made countless ideas feasible, and industries like law, education, and healthcare that historically didn't buy much software will transform over the next five years.
We can do bigger things with smaller teams, reach profitability with less capital, and build higher-quality products. While predicting 5-10 years out is difficult, at least for now and the next few years, this will be an unprecedented era of opportunity for entrepreneurs.
Change always brings fear. But looking at history, humanity has always found ways to move toward a better future. This AI revolution will be no different. The important thing isn't to fear change, but to understand and prepare for it. And now is the perfect time to start that preparation.