Most writing advice tells you to start with an outline. Build the skeleton, then add the flesh. Plan your ending before you write your beginning.
That advice works for some people. But for many first-time novelists, it's exactly backwards -- and it's why they get stuck at thirty thousand words.
The problem almost always traces back to the same place: the day they started writing, they didn't have a seed. They had fog.
"I want to write a really cool story" is not a seed. "I have a complete world built" is not a seed either. A seed is smaller, more primal, stickier than any of that. J.K. Rowling started Harry Potter with a single image: a boy who didn't know he was a wizard. No seven-book architecture. No magic system. Just one image that stuck to her brain for four hours on a train, making her ask: How does this boy discover the truth? What happens to the world when he does?
That's a seed -- a starting point that won't stop generating questions.
A Seed Is Not a Plot -- It's a "What If"
Too many people confuse "story seed" with "story outline." An outline is a blueprint. A seed is the thing underground that hasn't sprouted yet. It might be a question. It might be an image. It might just be a nameless ache in the chest.
Mark Twain's The Prince and the Pauper -- the seed was a hypothesis: what if a king and a beggar swapped identities? Hemingway's The Snows of Kilimanjaro -- the seed was a question: what does a person actually think about before dying? One situational, one emotional. Two wildly different seeds, both grew into classics.
How do you know if the seed in your hand is viable?
See whether it generates questions on its own.
"A person gets lost in a strange city" -- that's an event. No tension. But rephrase it: "A person gets lost in a strange city and discovers the streets are laid out exactly like a recurring dream -- and every road leads to the same address."
Reading that, your brain immediately fires off questions. What's at that address? Is the dream a prophecy or a memory? What happens when they arrive?
Good seeds do this. They don't need you to push. They sprout on their own -- a "what if" in the shower, another before bed, three more on the commute. That stickiness is the best quality indicator.
An idea that's been sitting in your head for three days without growing? It might just be a passing thought. Not a seed yet. Wait, or try another.
Where Seeds Come From
Three directions. Pick whichever comes most naturally.
Source One: Transforming Personal Experience
Life is the largest material library there is. But writing experiences directly as stories produces boring results about nine times out of ten.
The key isn't the event itself. It's the emotional core underneath.
In 2018, Celeste Ng mentioned in an interview that the seed for Little Fires Everywhere came from the anxiety she felt while moving -- not the act of moving, but the question "Where do I actually belong?" She dropped that anxiety into a completely different setting: two families, one small town, a conflict about belonging.
The method is simple. Recall the event from the past three months that shook your emotions the hardest. Not the event -- the emotion. Fear? Anger? Loss? Pull it out like a loose thread.
Then ask: what kind of story could this thread be stitched into?
The anxiety of missing a flight could become "a person who misses their flight by five minutes, thereby dodging a disaster, and starts believing they can predict the future -- until the predictions spin out of control."
See that? The story isn't about a flight anymore. It's about control. Fate. How people handle uncertainty.
Source Two: Pure Hypothesis
Take an everyday scene and throw in an unusual variable. Sci-fi and fantasy writers love this move, but it works for any genre.
A few examples:
- What if memories could be bought and sold, and what would the black market price be?
- What if divorce required both people to complete a task together before it became legal?
- What if the afterlife had a customer service center, and the line never moved?
Good "what ifs" chain-react. Memories can be traded -- who would sell their childhood? Would someone steal another person's first-love memories? How would the law handle memory copyright? Each follow-up question is story fuel, and they keep coming whether you want them to or not.
Try it. Set a timer for five minutes and write ten "what ifs." No filtering. The more absurd, the better. When you're done, look at which one makes you most desperate to keep asking.
That one is a candidate seed.
Source Three: Starting from Theme
Sometimes what's circling in your head isn't a concrete scene but an abstract idea. "Technology is changing intimacy." "Why do people in middle age suddenly doubt they took the wrong path?" "Can trauma be inherited?"
Abstract concepts can't be planted directly as seeds. They need a vessel -- a character who gets backed into a corner by the question.
Writing about "missed choices"? Then you need a specific person. Someone who gave up their dream of music to become an accountant thirty years ago, and one evening hears a stranger on a street corner playing the melody they once composed.
Suddenly "missed choices" isn't abstract anymore. It has a face, a sound, and the potential to make a reader's chest tighten.
Ask yourself: what question keeps circling your mind lately? What kind of character would be pushed to the edge by that question? The moment character and question collide, the seed takes shape.
Is This Seed Worth a Year of Your Life?
Not every idea can carry a novel. Too many writers get hooked by a "cool concept," bury themselves in eighty thousand words, then realize -- the story was done by chapter three.
Before you commit, run the seed through three gates.
The Elevator Test
Imagine meeting a stranger in an elevator. Thirty seconds. Can you explain the story?
Format: "This is a story about [someone] who must [do something] or else [consequence]."
Can't articulate it? That doesn't mean the seed is bad. It means it needs more crystallization. Fuzzy seeds grow fuzzy stories.
The Complexity Test
Can this seed hold a hundred thousand words?
Long novels need multiple conflict lines, character growth arcs, at least two or three subplots. If a seed's story can be told in five thousand words, it's a short story. Stretching it only dilutes -- like adding a liter of water to a shot of espresso.
Short stories are great. Write the short version first. Nail the core. If the root system turns out deeper than expected, expand later.
The Passion Test
This gate is the cruelest and the most honest.
Writing a novel is a marathon. A year. Maybe longer. There will be moments you want to quit -- not one, many. The only thing that gets you through is genuinely caring about this story.
The right seed will drag you out of bed at 3 AM because you suddenly thought of what a character should say in a particular scene. It'll barge into your head while you're cooking, walking, zoning out. Uninvited.
One more question worth sitting with: is this a story only you can write? Not that the idea needs to be unprecedented -- most stories have been told. But your experiences, your perspective, your emotional connection to this subject -- can they bring something nobody else can offer?
If the answer makes your chest warm, that's the one.
Using AI to Help Seeds Sprout
AI won't think up your story. That part is yours alone.
But AI can help you test whether a seed has potential much faster, saving you three days of staring at the ceiling. In Slima's AI Assistant, three approaches work particularly well.
Method One: Generate Variants
Say the seed is "a person who can hear plants talking." Open the Chat Panel (Cmd/Ctrl + J) and type:
I have a story seed: "A person who can hear plants talking."
Generate 5 different variants, each with a different character background and story direction.
AI might throw out these directions: a botanist who hears an entire forest wailing, a city office worker who discovers a forgotten garden in a basement, or -- this is a curse: if you hear a plant begging for help and don't act, you permanently lose one of your senses.
Don't accept everything. Pluck the elements you like and remix them into your own version. AI provides ingredients. You're the chef.
Method Two: Deep Questioning
Have AI play a ruthless editor and interrogate your seed:
My story seed is "an amnesiac detective."
Ask me 10 questions about this seed to help me think about the story's depth.
Don't give answers, just ask questions.
AI might ask: How did the memory disappear -- accident, disease, or sabotage? Before the amnesia, was this detective a good person or a bad one? Is someone waiting for them to remember? Does that someone hope they remember everything, or hope they never do?
These questions don't need answers right now. But they'll land like stones dropped into a pond, ripples spreading outward, showing you how deep the root system beneath the seed actually runs.
Method Three: Scene Testing
The fastest seed-validation method. Describe a key scene and have AI expand it:
Write a scene: the protagonist hears plants talking for the first time.
About 200 words, with specific sensory details.
After reading it, pay attention to your body. Heart rate up? Fingers itching? Wanting to open the editor and keep going?
If yes -- congratulations. This seed is alive.
If you feel "meh" -- no problem. Move to the next one. Testing seeds costs almost nothing. Finding the right one is the goal.
Practice in Slima
Theory's done. Open Slima. Follow along.
Step One: Free Writing (10 minutes)
Create a new file in Writing Studio. Name it "Story Seeds."
Press Cmd+D to enter Zen Mode -- fullscreen, zero distractions. Set a timer for ten minutes, then start writing. Any story idea that surfaces counts. Full sentences fine, three keywords also fine. Don't judge, don't go back, just push forward.
Target: at least twenty ideas. Can't reach twenty? Force it. The best ideas often hide right past the moment you think you've run dry.
Step Two: Mark (5 minutes)
Scan back through. Add ** next to the ideas that make your pulse tick up slightly.
The standard here is "want," not "should." "This topic has a big market" is not a reason. "This idea won't stop buzzing in my head" is.
Step Three: Elevator Test (10 minutes)
Pick three to five marked ideas and try a one-sentence summary for each:
"This is a story about [someone] who must [do something] or else [consequence]."
Stuck? Open AI Assistant (Cmd/Ctrl + J), paste the idea in:
My story idea is "a person who can hear plants talking."
Help me summarize it in this format:
"This is a story about [someone] who must [do something] or else [consequence]."
Give me three different versions.
AI's versions won't be perfect, but they'll show you different facets of the seed. Sometimes the best summary comes from tearing two of AI's versions apart and reassembling the pieces.
Step Four: Choose Your Seed
Look at those one-sentence summaries. Which one makes you most want to keep digging?
That's your seed.
Save it as its own file. Use the File Tree to create a dedicated folder if you want, or just drop it in the root directory. In the rest of this series, we'll take this seed and grow it step by step -- characters, structure, a complete story.
A seed doesn't need to be perfect. Doesn't need to be complete. It just needs to make you unable to stop.
Next up, we start from the seed and build a protagonist readers actually care about.
Now -- write the seed down.