Story Intelligence Infrastructure
Every script deserves
a real read.
Scripitt reads screenplays, pilots, pitches and manuscripts end to end and returns evidence-based evaluation in minutes — a verdict, a weighted scorecard and a Decision Brief your team can act on.
AI reads. Humans decide.
Sample report · excerpt
COLDWATER
Marin Held · Drama · Script Coverage
Verdict
Recommend
Alignment Radar · best lanes
Top strength. The premise and the payoff are the same object. The warm pool Nora finds in the first scene ('It's warm.') is explained by Dean's confession ('It has to be Amara doing two lengths. It has to be something after'), and the image is resolved when Nora puts her legs into that water. This gives the short unusual unity and a strong visual identity.
Read the full report →
Int. development office — night
The pile is bigger than the week. So the pile decides what gets read, and the reading decides what gets made.
Days
Turnaround on a single evaluation, when a human reader has a queue.
Inconsistent
Two readers, two rubrics, two verdicts on the same script — and no way to reconcile them.
Unread
The submissions nobody got to. The reason they were passed on is that nobody opened them.
Scripitt is the intelligence layer underneath that problem. It does not replace the person who decides. It makes sure that person is deciding on a complete, consistent read of everything in front of them.
Decision architecture
One read. Four connected layers.
Depth when you need to understand. Compression when you need to decide.
01
Analysis
What does the evidence say?
The full Story Intelligence Report — deep analysis, scoring, development, alignment, market, adaptation and risk intelligence, with citations back to the page. For the people who need the evidence.
See the evidence →02
Recommendation
What does Scripitt conclude?
The Executive Decision Brief — the recommendation, rationale, confidence, critical opportunity, material risk, alignment, next action and decision-changing factors, derived from the same record. For the person who has to make the call.
Make the call →03
Human decision
What did the organization actually decide?
The Company Decision — Recommend, Consider or Pass, by whom, on which draft, when, and why — recorded as its own state, separate from and above the machine recommendation.
04
Memory
What should the organization retain?
The library keeps the analysis, the recommendation, the human decision, the material history and every revision connected to one intelligence record, so the organization learns from its own reads over time.
Five capabilities
- 01Read & Understand
- 02Evaluate & Score
- 03Route & Compare
- 04Remember & Organize
- 05Operationalize Insights
Modes
Eleven modes, eleven kinds of read
A pilot is not a feature, a pitch is not a pilot, and a manuscript is none of them. Each mode carries its own rubric, its own sections and its own idea of what “working” means, and the mode is detected on the way in.
01
Feature Script
Full coverage on a feature screenplay — logline, synopsis, craft breakdown, market read and a Pass / Consider / Recommend verdict.
02
TV Pilot
Pilot-specific analysis: engine, series-ability, character runway, episodic vs. serialized fit, and where the season two problem lives.
03
Book & Manuscript Review
Long-form editorial coverage for publishers and literary teams — structure, voice, market position and comparable titles.
04
Quick Book Triage
A fast first-pass read for a full submission pile. Sort the stack before anyone spends a weekend on it.
05
Book-to-Screen Adaptation
Adaptation feasibility: what survives the translation, what has to be invented, budget shape and the format the material actually wants.
06
Publishing Readiness
Submission-grade assessment of whether a manuscript is ready to go out, and precisely what is holding it back.
07
Micro-Drama & Vertical Series
Built for the vertical short-form format on its own terms: hook velocity, per-episode cliff placement, retention across an episode arc, and whether the premise sustains a full season at that length.
08
Topical Coverage
Analyzes time-sensitive material against current web sources to identify relevant developments, missed topical opportunities, and story context, with citations provided for human review. The only mode that researches the live web; the other ten read the submitted material alone.
09
Pitch Document
Scored as a pitch, not a script: premise and hook, audience and positioning, character and world, stakes and dramatic engine, series or film potential, and how persuasively it reads. Never marked down for missing sluglines, dialogue or act breaks.
10
Series Treatment
Evaluates a season on the page: protagonist and ensemble arcs, the episodic and season engine, progression and stakes, tone and theme. Screenplay structure and formatting are not evaluated and never deducted.
11
Film Treatment
Reads a feature as prose: the protagonist's arc, conflict and stakes, beginning-middle-end progression and cinematic potential, on its own 100-point rubric rather than a screenplay's.
Inside the platform
Evaluation is the output. The judgment around it is the product.
Evaluation Lens
Patent pendingCoverage is never generic. Choose the lane, the buyer and the creative sensibility the material is being judged against, and the read changes accordingly.
Trainable taste
Patent pendingCalibrate a lens from the scripts an executive actually loved, or simply describe the taste in a few sentences. The system learns that sensibility and applies it consistently across the pile — and it cannot be instructed to force a verdict.
Alignment Radar
Patent pendingNames the lanes the material fits, the buyers most likely to respond and the kindred voices it sits beside — each with a fit score, not a vibe.
What-If Simulator
Patent pendingRe-score the material under an as-executed assumption. Separates what is wrong on the page from what is merely unproven.
Decision Brief
Patent pendingA decision-grade executive layer derived from the full Story Intelligence analysis. It preserves the recommendation, rationale, confidence, critical opportunity, material risk, alignment, next action and decision-changing factors — while the complete supporting analysis remains available underneath. Shareable by link, exportable to DOCX, carrying your letterhead.
Company Decision
Patent pendingThe AI recommends; the organization decides — and the platform records which is which. A verdict, who made it, when and why is logged against the project and travels with the brief and the library register, so the human call is never confused with the machine's.
Side-by-side
Put two or three projects against each other and get a head-of-development recommendation with the reasoning shown.
Ask the library
Patent pendingEvery report is indexed. Ask your own catalog a question in plain language and get answers with the coverage cited.
Provenance Ledger
Patent pendingEach submission is fingerprinted at intake and written to an append-only chronology, with the coverage number bound to that exact revision. A Creator Provenance Report can be exported for any work, stating what is known and naming what is not.
White-label
Patent pendingEnterprise workspaces carry your logo, your colors and your letterhead across the workspace, every shared report and brief, the preview card a recipient sees, and the DOCX export. Brand identity is imported from your website in one step. Nothing on an outward surface says Scripitt unless you want it to; every white-labelled surface carries a discreet Powered by HAAIS mark.
Library registers
Export or print a register of the desk, the cloud library or server activity — material and coverage numbers, recommendation, the recorded company decision, what was shared and when. Metadata only; the text never leaves.
What is actually running
An engine, not a chatbot.
Scripitt is a story-intelligence system. A language model is one component inside it, called at defined steps to read and assess. Everything around those steps is Scripitt's own engineering, and none of it can be talked into a different answer.
Change the model and the platform keeps working. Remove the platform and the model is just a prompt.
Intake and structure
Eight formats, scanned pages, size-aware routing and a complete comprehension pass that structures the whole work before any judgment is formed.
Orchestration
Mode-specific specialist reads run in parallel against the lens you chose, then a reconciliation step resolves their disagreements instead of averaging them away.
Validation
Every output passes a bounded validation ladder that rejects, repairs and re-checks weak work. A report that cannot clear it says so rather than inventing confidence.
Scoring and verdict
Weighted scorecards, strategic scores and the Pass / Consider / Recommend verdict are computed by the system from the evidence, not improvised in prose.
Memory and provenance
Fingerprints, an append-only chronology, same-text detection, revision-aware scoring and coverage numbers bound to the exact draft.
The human layer
The company decision is recorded as its own state, above the machine recommendation, with who, when, which draft and why.
Where we stand
The machine reads.
You decide.
This is not a rhetorical position. It is how the system is built, and it is the reason a writer can hand us a draft without hesitating.
The machine never writes
Scripitt does not generate, rewrite or punch up creative material. Not a line. It reads, it assesses, it cites. Authorship stays entirely with the writer.
The decision stays human
Every output is an input to a person's judgment. The verdict is a recommendation with its reasoning exposed, so a reader can disagree with it on the evidence.
Nothing is asserted without the page
Claims trace back to the material. If the read cannot support a judgment, the report says so rather than inventing confidence.
Built for
Production companies
Clear the pile without losing the good one at the bottom of it.
Managers & agencies
Read every client submission properly, at the speed clients expect.
Publishers
Triage the slush, then go deep on what survives.
Contests & labs
Consistent, defensible scoring across thousands of entries.
Vertical & micro-drama
Volume formats need volume reading. This is what it was built for.
Pilots are open.
We are taking on a small number of production companies, agencies and publishers to run Scripitt against a live slate. Tell us what you read and how much of it, and we will come back with a scope.