Loglines, Synopses & Treatments

Here's a truth that took me too long to accept: your screenplay's first three readers never read the screenplay. They read a sentence, and if the sentence works, a page, and if the page works, maybe ten pages of prose. Only then does anyone open the script. I call this the paperwork trilogy — logline, synopsis, treatment — and it's the most compression-heavy, least glamorous writing in the business. It's also the development job where AI assistance pays off fastest, because compression is a mechanical skill and machines are good at mechanics. The catch comes at the end, and it's a big one. Hold that thought.

First, know the three documents cold:

• Logline — one or two sentences (about 25–50 words). Protagonist, want, obstacle, stakes. Its only job is to make someone say "tell me more."
• Synopsis — half a page to one page. The whole story including the ending, in clean present-tense prose. Its job is to prove the premise actually becomes a movie.
• Treatment — roughly 3 to 10 pages for a feature. The film told as a story: acts, major sequences, key turns, the emotional ride. Its job is to make a reader feel like they watched the movie.

The compression workflow

The workflow runs on one principle: your outline is the source; AI is the compressor. Feed your full outline into the script's project — the one we built in Chapter 3 — and ask for each document at its target length. Ask for five loglines, not one, emphasizing different angles: the relationship, the ticking clock, the irony. Ask for a one-page synopsis, then interrogate it — what did it cut that you consider essential? Its choices about what survives compression are a diagnostic in themselves; when the machine drops a subplot without the story collapsing, that's worth knowing before a producer notices the same thing.

Loglines get a special test I use on every project: make the AI pitch it back. Give it your logline cold — in a fresh conversation with no context — and ask: "Based only on this logline, what movie do you imagine? What genre, what tone, what's the poster?" If the machine imagines a different film than yours, so will everyone in the room. That's a fixable problem, and it's much cheaper to fix at the logline stage than in a pitch meeting.

Since this chapter rides on everyone hunting for a screenplay treatment example, here's the structural pattern of a working treatment paragraph — one story beat, told like the movie feels:

MAYA, 34, repossesses cars for the bank that repossessed her childhood home — she's good at it, and she hates that she's good at it. When a routine job puts her late father's Impala on her list, sold twice since the foreclosure, she breaks her own first rule: never look in the glove box. Inside is a letter in her father's handwriting, addressed to her, dated the week he died. She doesn't return the car. Now the bank is hunting its own repo agent, and Maya is driving stolen property across three states to find out why her father knew she'd come looking.

Look at the mechanics: character introduced with a contradiction, not a résumé. Present tense. A turn inside the paragraph. Concrete images — the glove box, the handwriting — instead of summary phrases like "she discovers a secret." And it ends leaning forward into the next beat. That's a treatment doing its one job: reading like the movie feels, not like a book report about it. String fifteen to twenty-five of those paragraphs across three acts and you have a feature treatment.

The catch: these documents sell voice

Now the warning I promised. AI will produce a clean draft of all three documents, and the treatment draft will be smooth, professional — and voiceless. Remember Chapter 4: the machine converges on the middle. But here's what makes this trilogy different from a placeholder scene: a reader takes the paperwork as a sample of your writing. A logline, a synopsis, a treatment — these are auditions. If the prose is generic, the reader concludes the script is generic, and never opens it to find out they're wrong.

So the rule is absolute: the AI draft is never the final draft. Use the machine for structure — what survives compression, what order the beats land in, where the act breaks fall in prose. Then rewrite every sentence in your own hand, the way you'd punch up dialogue: sharper verbs, your rhythm, the specific details only you know. The compression is mechanical and delegable. The performance is not. In my experience the rewrite takes a third of the time writing from scratch would have — the machine did the arithmetic; you do the music.

You've got a premise, and now the paperwork to pitch it. Next chapter we build the load-bearing document underneath all of it — the outline — and how to use AI for structure passes without letting it flatten your story into formula.