Software is literature now.
I'm not being metaphorical. Your IDE is a chat window. Your compiler is a conversationalist. Every interaction with AI is an act of writing. We're not coding anymore - we're conducting conversations that become capabilities.
The Compression
Here's what's coming: Every piece of software will compress down to ~100 lines of agent loops calling a foundation model. Your entire app - the one you spent years building - will fit in a GitHub gist. When that happens, what differentiates products? Not the code. Not the infrastructure. The narrative. The taste. The imagination.
Code isn't the thing anymore. Working with AI is. And working with AI is writing.
Why Fiction, Though?
Remember that quote from Antoine de Saint-Exupéry? "If you want to build a ship, don't drum up people to collect wood and don't assign them tasks and work, but rather teach them to long for the endless immensity of the sea." Atomic Habits teaches you about wood and tasks. One Hundred Years of Solitude teaches you about the sea. Here's what fiction gives you that documentation doesn't:
- *Understanding Intent vs Expression*: Humans rarely say what they mean. In Pride and Prejudice, nobody says "I love you" for 200 pages, but you know exactly who loves whom. That's prompt engineering - reading between the lines of what users ask for versus what they need.
- *Edge Case Imagination*: Science fiction writers have been prompt engineering since before computers existed. Asimov's Three Laws of Robotics? That's literally alignment research. Every sci-fi story is a unit test for the future.
- *Pattern Recognition Beyond Data*: Fiction reveals emotional patterns that your telemetry will never capture. A user journey isn't a funnel - it's a hero's journey, complete with resistance, transformation, and return.
The Concrete Skills (For You Engineers Who Need ROI)
Fine, let me spell it out:
- RAG Systems → Read Borges' "The Library of Babel." Your retrieval system is navigating infinite possibility space. Until you understand that philosophically, you're just building search.
Speaking of Borges - you know "Funes the Memorious"? The man who remembers everything but understands nothing? That's your RAG system without wisdom. Perfect recall, zero comprehension. Engineers who've read Funes understand why 100% recall isn't the goal. They build systems that know how to forget, how to abstract, how to see forests instead of cataloging every leaf.
- Chain-of-Thought Prompting → Detective fiction. Sherlock Holmes doesn't just solve crimes - he shows his work. That's literally what CoT is.
- Fine-tuning → Read "Pygmalion" or watch "My Fair Lady." You're not training a model; you're teaching it to speak like the natives of your domain.
- Prompt Engineering → Jane Austen. Every prompt is a social negotiation where what's unsaid matters more than what's said.
- AI Alignment → Start with Asimov, graduate to Ted Chiang, PhD in Greg Egan. The technical papers will make sense after you've felt the problems in your bones.
The New Engineering
You know what's wild? The engineers making bank at AI companies aren't pure wordcels or pure shape rotators. They're hybrids. They can write prompts that make AI sing AND debug why it's eating 400GB of RAM. They read Borges for inspiration, then optimize the vector database. They understand narrative structure AND distributed systems. The pure wordcels are writing beautiful prompts that crash at scale. The pure shape rotators are building bulletproof systems nobody wants to use. The winners live in both worlds.
Think about what actually happens now. You prompt an AI, it writes the code. But someone still needs to know if that code will survive production. Someone still needs to architect systems that won't collapse when they succeed. Every great AI product needs both: the imagination to see what's possible and the engineering to make it real. The 100-line future isn't about code disappearing - it's about code becoming poetry that actually runs.
Permission to Read
I know what you're thinking. "I don't have time for fiction. I need to keep up with the latest papers."
Listen: Reading fiction IS keeping up. It's the highest-leverage activity you can do right now. Tell your manager you're developing critical AI engineering skills. You're not lying.
Start here:
- This week: Any Borges short story (they're like 5 pages, you have no excuse)
- This month: A detective novel for chain-of-thought practice
- This quarter: One ambitious sci-fi novel that breaks your brain a little
We've massively accelerated the speed of building. Now it has become important to stop and ask what we're building. And we must all build a maximally pleasant future for ourselves and the future human race. The future belongs to engineers who can imagine it. And imagination, like any muscle, needs exercise.
The Market Already Knows
Look at what companies are actually hiring for:
[AI Prompt Engineer - Token Metrics] - They want a "Bachelor's degree in Computer Science, AI, Journalism, or related field." Journalism! They're literally saying storytellers welcome.
[Japanese Prompt Engineer - Welocalize] - Requires creating "narrative stories" and "deep understanding of intricate relationships within narratives." Not code. Stories.
[AI Writer positions] - 800+ openings, $28-$96/hour. The job description? "Not just a writer — you're a storyteller, strategist, and trend translator."
[Conversational Designer roles] - 200+ jobs, up to $230k. Because someone needs to design how AI talks to humans, and that someone better understand dialogue.
The job titles are evolving before our eyes. "Prompt Engineer" is just the beginning. Companies are realizing they need people who can think in narratives, not just nested loops.
Software is literature now. The question is: are you reading enough to write the future?
What story is your code telling? What fiction has shaped how you think about building? Drop me a line - I'm always looking for book recommendations that connect to this new world we're building.
And now on to the jobs
I'm not being metaphorical. Your IDE is a chat window. Your compiler is a conversationalist. Every interaction with AI is an act of writing. We're not coding anymore - we're conducting conversations that become capabilities.
The Compression
Here's what's coming: Every piece of software will compress down to ~100 lines of agent loops calling a foundation model. Your entire app - the one you spent years building - will fit in a GitHub gist. When that happens, what differentiates products? Not the code. Not the infrastructure. The narrative. The taste. The imagination.
Code isn't the thing anymore. Working with AI is. And working with AI is writing.
Why Fiction, Though?
Remember that quote from Antoine de Saint-Exupéry? "If you want to build a ship, don't drum up people to collect wood and don't assign them tasks and work, but rather teach them to long for the endless immensity of the sea." Atomic Habits teaches you about wood and tasks. One Hundred Years of Solitude teaches you about the sea. Here's what fiction gives you that documentation doesn't:
- *Understanding Intent vs Expression*: Humans rarely say what they mean. In Pride and Prejudice, nobody says "I love you" for 200 pages, but you know exactly who loves whom. That's prompt engineering - reading between the lines of what users ask for versus what they need.
- *Edge Case Imagination*: Science fiction writers have been prompt engineering since before computers existed. Asimov's Three Laws of Robotics? That's literally alignment research. Every sci-fi story is a unit test for the future.
- *Pattern Recognition Beyond Data*: Fiction reveals emotional patterns that your telemetry will never capture. A user journey isn't a funnel - it's a hero's journey, complete with resistance, transformation, and return.
The Concrete Skills (For You Engineers Who Need ROI)
Fine, let me spell it out:
- RAG Systems → Read Borges' "The Library of Babel." Your retrieval system is navigating infinite possibility space. Until you understand that philosophically, you're just building search.
Speaking of Borges - you know "Funes the Memorious"? The man who remembers everything but understands nothing? That's your RAG system without wisdom. Perfect recall, zero comprehension. Engineers who've read Funes understand why 100% recall isn't the goal. They build systems that know how to forget, how to abstract, how to see forests instead of cataloging every leaf.
- Chain-of-Thought Prompting → Detective fiction. Sherlock Holmes doesn't just solve crimes - he shows his work. That's literally what CoT is.
- Fine-tuning → Read "Pygmalion" or watch "My Fair Lady." You're not training a model; you're teaching it to speak like the natives of your domain.
- Prompt Engineering → Jane Austen. Every prompt is a social negotiation where what's unsaid matters more than what's said.
- AI Alignment → Start with Asimov, graduate to Ted Chiang, PhD in Greg Egan. The technical papers will make sense after you've felt the problems in your bones.
The New Engineering
You know what's wild? The engineers making bank at AI companies aren't pure wordcels or pure shape rotators. They're hybrids. They can write prompts that make AI sing AND debug why it's eating 400GB of RAM. They read Borges for inspiration, then optimize the vector database. They understand narrative structure AND distributed systems. The pure wordcels are writing beautiful prompts that crash at scale. The pure shape rotators are building bulletproof systems nobody wants to use. The winners live in both worlds.
Think about what actually happens now. You prompt an AI, it writes the code. But someone still needs to know if that code will survive production. Someone still needs to architect systems that won't collapse when they succeed. Every great AI product needs both: the imagination to see what's possible and the engineering to make it real. The 100-line future isn't about code disappearing - it's about code becoming poetry that actually runs.
Permission to Read
I know what you're thinking. "I don't have time for fiction. I need to keep up with the latest papers."
Listen: Reading fiction IS keeping up. It's the highest-leverage activity you can do right now. Tell your manager you're developing critical AI engineering skills. You're not lying.
Start here:
- This week: Any Borges short story (they're like 5 pages, you have no excuse)
- This month: A detective novel for chain-of-thought practice
- This quarter: One ambitious sci-fi novel that breaks your brain a little
We've massively accelerated the speed of building. Now it has become important to stop and ask what we're building. And we must all build a maximally pleasant future for ourselves and the future human race. The future belongs to engineers who can imagine it. And imagination, like any muscle, needs exercise.
The Market Already Knows
Look at what companies are actually hiring for:
[AI Prompt Engineer - Token Metrics] - They want a "Bachelor's degree in Computer Science, AI, Journalism, or related field." Journalism! They're literally saying storytellers welcome.
[Japanese Prompt Engineer - Welocalize] - Requires creating "narrative stories" and "deep understanding of intricate relationships within narratives." Not code. Stories.
[AI Writer positions] - 800+ openings, $28-$96/hour. The job description? "Not just a writer — you're a storyteller, strategist, and trend translator."
[Conversational Designer roles] - 200+ jobs, up to $230k. Because someone needs to design how AI talks to humans, and that someone better understand dialogue.
The job titles are evolving before our eyes. "Prompt Engineer" is just the beginning. Companies are realizing they need people who can think in narratives, not just nested loops.
Software is literature now. The question is: are you reading enough to write the future?
What story is your code telling? What fiction has shaped how you think about building? Drop me a line - I'm always looking for book recommendations that connect to this new world we're building.
And now on to the jobs