Quick definition
Instead of one general-purpose AI giving vague notes, this method assigns specific specialist roles — structure, dialogue, punch-up, action, theme, continuity — and runs each of them on a script in a deliberate order, the way a real writers' room distributes different kinds of attention. Like a tabletop party that needs a balanced mix of roles rather than five fighters, a script benefits from several narrow, well-prompted specialists more than one broad, vaguely-prompted generalist.
In one line: Fresh eyes helped in the room for a reason — now you can run a whole room of them.
Why "make it better" prompts produce nothing useful
A vague instruction gets a vague response — "make it better" produces generic suggestions with no diagnostic value. A specific, genre- and reference-anchored prompt ("look at this pilot through the Alias lens — teaser that grabs you by the throat, four acts that escalate") gives the model a concrete standard to measure against, which is what actually produces a usable note instead of noise.
Six specialist roles at a glance
| Specialist | Focus | Sample prompt direction |
|---|---|---|
| Structure | Sequences, acts, pacing | "Where does the second act drag like a late-season bottle episode?" |
| Dialogue | Line-level polish and voice | "Give me three versions of this argument in different registers." |
| Punch-up | Humor as tension release | "Find places where character-based humor could break tension without killing momentum." |
| Action | Visceral but readable staging | "Make this fight feel visceral but stay readable." |
| Theme | What the story is actually about | "What's the central tension driving this story?" |
| Continuity | Plot holes, tonal shifts | "Be the network executive who actually read the script." |
Order matters — fix structure before polish
Running specialists in sequence — structure first, then dialogue, then continuity as a final pass — mirrors how a real rewrite has to work: a structure problem buried under great dialogue is still a structure problem. Polishing lines before the underlying scene order is sound wastes the polish, since a fix at the structural level frequently invalidates dialogue built on top of it.
Treat the notes like network notes, not verdicts
Some AI notes will be useful, most won't, and all of them are worth thinking about — the same triage a writer already applies to human executive notes. The value isn't that every specialist bot is right; it's that being wrong in an interesting way can still crack open a problem, the same function an intentionally bananas pitch can serve in a live writers' room to unstick a stuck conversation.
The showrunner is still the showrunner
These tools are useful for challenging instincts and forcing a writer to either defend a choice or recognize when they're being lazy — but the final call on whether a joke lands, a reveal is earned, or a character would actually say a given line still belongs to the writer. AI specialists are best understood as a room of objective reviewers with infinite patience, not a replacement for the judgment that decides which of their notes to take.
AI-specialist-room checklist
- Is my prompt specific and reference-anchored, or a vague "make it better"?
- Am I running structure notes before dialogue polish, in that order?
- Am I treating each note as a data point to weigh, not a verdict to obey?
- Is there a note here that's "wrong" in a way that's actually pointing at something real?
Try it on your own script
Scribbler’s Scene Doctor & Character Doctor
Use the Scene Doctor as a structure-and-scene specialist, checking objectives, escalation, and change by the end of a scene. Use the Character Doctor as a character-and-voice specialist, checking whether traits and behavior read clearly.
Analyze a scene → · Analyze a character →
Source note
This maintained reference page develops ideas first published in “Virtual Script Doctors?” in Scribbler’s Toolbox on February 9, 2026. The original essay remains part of the editorial archive; this version is structured as a durable craft reference.
Related pages
- How to Vibe Code an AI Writing Tool: Building Scene Doctor — the origin story of one of these specialists, built as a working tool rather than a prompting technique.
- The WANT Framework for Screenwriting: How to Give Every Character a Playable Scene Objective — the kind of specific methodology a structure specialist should be prompted with, rather than generic craft advice.
About the author
Jesse Alexander is a screenwriter, showrunner, and narrative designer whose work spans film, television, games, and transmedia storytelling. Scribbler’s Toolbox collects practical frameworks developed from professional story development and production.