EtherPen is an AI script engine for short-drama writers and production studios: it generates full episode-by-episode scripts from an outline, runs five-dimension reviews with pinpoint revisions on existing drafts, and adapts Chinese hit scripts into localized shows for overseas markets.
Editor's synopsis · The revenge throughline holds and the heroine's arc is clear; pay points drift off the act climaxes in eps 12 and 24, and hooks run weak in eps 3–7 — tighten per the model fixes…
Video models went through three paradigm shifts in 18 months; good scripts stay scarce through all of them. We focus on the layer that makes the video worth shooting.
Every AI change ships as a tracked-changes preview a human accepts before it lands; untouched lines stay verbatim.
AI automates what can be measured; senior editors focus on judgment — 200 episodes of screening compressed into 50 that matter.
One engine, one script, two interfaces. Switch any time — the draft stays put.
Write like it's Word. Edit episode by episode, annotate a selection, keep every revision on the record, with the whole structure open beside you.
Say what you want; it handles the rest. No need to know act structure, no tool to learn.
Write a new one, fix an old one, take a hit global. All three share one data contract, so the draft you review today is the one you keep writing tomorrow.
Writers do exactly three things: submit the outline, approve the master sheet, green-light episodes card by card. Behind the scenes, AI builds the character system, act structure, and paywall placement — cutting every pay point on the eve of an act climax.
Bring any asset — an outline, a synopsis, a half-finished draft, or a full script. Get a five-dimension review report in about 9 minutes, fix issues point by point on a Word-style tracked-changes desk, and hand off to the creation line whenever you're ready.
Three confirmations — genre, market, reference show. Then skeleton inheritance plus reskin assembly: emotional anchors and pay-point structure are replicated episode by episode, while characters, worldbuilding, and cultural hooks are rebuilt for the US, Brazil, or Indonesia.
General-purpose LLMs can write dialogue; they can't write paywall structure. Every generation, review, and adaptation in EtherPen is benchmarked against structural assets from real hits.
T1 positive samples, plus 37 flop negatives for discriminative validation
30 real overseas short dramas — 25,535 action lines / 28,643 dialogue lines
A dictionary from Chinese story elements to overseas emotional equivalents
26 English / 4 Portuguese / 4 Indonesian
Legally sourced and abstracted to pattern level — zero copyrighted text
Per-market cognition matrices for the US, Brazil, and Indonesia
The market is doubling — yet what to write, whether it holds, and how to go global still run on manual experience. That is the window for a script-layer engine.
2025 overseas in-app revenue of Chinese short-drama apps, on 2.15B downloads
Source: DataEye · KchuhaiWhether something is trending isn't the question — narrative portability is. Radar doesn't pile up data; it uses AI to tell you why something is hot and whether it can become a short drama.
Covers the US, Brazil, and Indonesia, updated daily. Once live, the Free tier includes the global Top 50 trend board.
Notify me at launchRadar data comes from compliant collection of public charts and official APIs, for creative reference only.
The main chart is built on tropes (archetype × relationship × conflict), updated weekly — not two dozen rigid genre buckets.
Per-show chart trajectories, surfacing breakout climbers at daily granularity.
Detects first-seen element combos that beat the baseline — "rebirth × 1980s × military marriage" — old bones in new skin, caught as it happens.
Every signal passes five checks (trope extraction / emotional promise / conflict compression / cultural portability / timeliness risk) into a 0–100 score.
Turning the methodology we run on — orthogonal act/paywall design, six-dimension adaptability, four-dimension craft, and the data behind what audiences pay for — into something teachable.
The Chinese edition of the academy is live. The English edition has not been built yet.
Go to the academy →The methodology itself is being written up, drawn from what actually runs in the product rather than from theory. English editions are not published yet.
Annual subscription, hard quotas, zero overage billing. Free runs on domestic models; paid tiers run on Claude deep models.
Individual creators · new writers
Independent writers · small studios
Short-drama studios · content labels
Top studios · custom needs
The invite-only beta is open. Tell us about your team and use case — we'll get you set up shortly.
Every shipped capability is free to use during the beta