Docs › Screenwriting › Plan a season before you write it: a beat board and a timeline

Plan a season before you write it: a beat board and a timeline

Last updated September 10, 2026 · 16 min read

In one line: laying out a season before you write it — a beat board for the shape, a timeline for the chronology, notes for what you plant and pay off, and an honest account of how loosely the three are wired to your episode files.

Story Skeleton.beats in Ink Tide (Screenplay). Act One (Setup) 5 · 4 · 3,858 words, Act Two (Confrontation) 6 · 6 · 6,127, Act Three (Resolution) 3 · 2 · 2,690 — the three numbers in a column header are beats, files and words. The twelve episodes are spread over fourteen beats, Episode 01 is expanded to its three scenes, and the beats nobody has written yet — Hesitation, Subplot, Final Image — hold their place with a Not written yet chip.

Two files, two different questions

A beat board (.beats)

  • Answers what shape
  • Columns are acts — consecutive stretches of the story
  • Order is story order
  • Holds your chapters and episodes as cards

A timeline (.timeline)

  • Answers when did this happen
  • Columns are labels — parallel threads running at the same moment
  • Order is in-story chronology
  • Holds events you write, not files

Most twelve-episode shows need the first one. You need the second as well the moment your story has a past — a case from a hundred years ago, a flashback structure, two threads that have to line up, or characters whose ages must add up.

The two files do not talk to each other. No import, no shared identifiers, nothing that generates one from the other. They are two documents you keep for two reasons.

The shape first: a beat board

Right-click in the file tree → More → New beat board. An empty board offers four ready-made structures, or Start blank:

Template Acts Beats
Three-Act Structure 3 14
Taiwanese TV (Three Acts, Six Beats) 3 6
Kishōtenketsu (Four-Part East Asian Form) 4 4
Freytag's Pyramid (Five Parts) 3 5

A template is a one-shot seed — it fills the board once and then gets out of the way. The picker only appears while the board is empty, so choose before you start dragging. Which to pick is in Choosing one of the four structure templates.

Two words to keep straight:

  • A beat is an intention: a title and a one-line summary. It points at nothing. "Midpoint — she learns the nine pages were torn out deliberately" is a beat whether or not anything is written yet.
  • A card is a file sitting on the board. That is all a card is.

Beats live in act columns; cards live inside beats, or loose in an act. Rename a column by clicking its name. Add one at the tail. Deleting a column keeps everything inside it — the beats and cards fall into an Unassigned column rather than into act one. Each column header carries three numbers: beats · files · words.

Hanging twelve episodes on it

Drag them in from the file tree. That is the only way — there is no picker, and no "new episode" button on the board. Only chapters and screenplays are accepted; anything else is refused with Only chapters and scripts can go on the board.

A file appears once on a board. Dragging it somewhere else moves it; it does not copy.

Act Two (Confrontation)
  Midpoint
    Episode 06 - The Drowned City      4 scenes
    Episode 07 - The Blank Stele       4 scenes  ▾
        1  INT.  The Blank Stele        night   ~2m / 5m
        2  EXT.  The Drowned City       dusk    ~1m / 2m
  Rising Pressure
    Episode 08 - Pei Zhao              4 scenes
  Reflection
    (Not written yet)
  • A screenplay card shows its scene count; a chapter card shows its word count. Both are recomputed live, so a card follows renames, moves and every scene you add.
  • Expand a screenplay card and you get one row per scene: number, interior/exterior, the location, the time of day, the estimated runtime and that scene's target minutes.
  • A beat with no cards wears a dashed border and a Not written yet chip. That is the whole "this isn't done" signal on the board.
  • Removing a card does not delete the file. Nor does deleting a beat delete its cards — they become loose cards in the same act. Both are undoable from the toast.

Those expanded scene rows write straight back into the episode. Rename a location, flip INT to EXT, change the time of day, set a target in minutes, reorder the scenes, even drag a scene from one episode into another — all from the board, all landing in the .script file. Slima takes an automatic checkpoint before the destructive ones. This is the most useful thing on the page and the easiest to miss.

On a phone the board is fully editable except for the one thing you need most: you cannot put a file on it. The only route in is a drag from the file tree, and there is no drag on a phone. Editing beats, reordering with the ▲▼ buttons and the Move to… sheet all work; adding an episode does not. Lay the season out on a desktop.

There is no progress bar, and that is deliberate

Say it plainly, because people look for it: a beat has no done state. There is no percentage, no "10 of 14 beats filled", no badge, no report. You get the Not written yet chip on empty beats and the three numbers in the column header, and you do the arithmetic — fourteen beats, twelve cards.

Letting the AI lay beats out

A beat board is one of the files the AI edits by field rather than by rewriting. Eight tools: add, rename, re-summarise and delete an act; add, retitle, re-summarise, delete and move a beat; and a batch that applies them together or not at all.

The AI cannot touch a file card. Not add one, not move one, not remove one — the tools do not exist. It has no reliable way to name a file in your book, so any card it "placed" would be an invention. Cards are yours, and they arrive by drag.

Its proposals appear on the board itself as ghost columns and ghost cards with dashed edges, each with Accept and Reject, plus a floating pill with the count and an accept-all. Deleting an act never deletes the beats inside it, and deleting a beat never deletes its cards — the same safety rails you have by hand. The general mechanism is in AI structural editing: why it edits by field and Reviewing the AI's structural changes in place.

Then the chronology: a timeline

Right-click → More → New Timeline. The first thing it asks is the only decision that matters:

Free ordering

  • No dates. Drag events into order
  • Write time however you like — three years later, the eve of the ban
  • Right when time is deliberately loose

Use a calendar

  • A calendar — the real one, or one you build for your world
  • Dated events sort themselves; a month-by-month view appears
  • Right for a hundred-year backstory or ages that must add up

You can change your mind at any time; turning the calendar on or off never touches your events or their order.

Then per event, three ways of expressing time, mixable in the same file: No time, Free text ("three years later"), or a Calendar date (which can be as vague as a year). Whatever you use, the order you dragged things into is the authority — the calendar is a way of displaying the line, not the thing that defines it.

  • Era is a free-text separator band for stretches of story time when you have no calendar: Before / Now, printed across the timeline as a divider.
  • Labels are the parallel threads — one per character or subplot. A label carries a colour and counts its own events; you can solo one, and when a thread goes quiet the timeline says so: "{name}: absent for the next N time points — time to wrap it up?" Events sharing a moment are marked Same time.
  • With a calendar on you also get a Calendar view alongside the timeline, at month or year scale.
  • There is an Export .timeline button in the toolbar; the file goes with you.
  • The timeline is fully editable on a phone — every interaction is a tap and a sheet.

Main Timeline.timeline in Ink Tide (Screenplay), switched to the Timeline view. The two columns are the two labels — Main with 9 events, The Past with 7 — and the counts sit in the headers. The rules across the width are bands the calendar draws by itself: Moon 1 Chengwen 103, then Moon 3 and Moon 4 of Chengwen 107. The Past stops at Chengwen 103 while Main only starts four years later, which is the gap you are meant to see. The selected card is Copying the Tide Records, Shen Yan sees a word move, Moon 3 12, Chengwen 107.

How much the timeline knows about your episodes: less than you think

Dragging an episode onto a timeline does not link it. You get a plain event named after the file — no jump-to-file, no back-reference, and renaming the episode afterwards will not rename the event.

The AI cannot link an event to anything either. It is blocked from setting a note, a scene or a file on an event, and it says so: entries it creates are always standalone.

There is no "build a timeline from my episodes". Nothing reads your screenplays and lays their scenes out in time.

So treat a timeline as a document about the story's chronology, written in your own words — which is what it is good at. If you want an event to point at an episode, say so in the event's title: "Ep 3 — the letter in the lining".

The one thing that really does cross episodes: setups and payoffs

Planted in episode 2, paid off in episode 8

Right-click the line in episode 2 Add note → Add Setup (Foreshadow) Link payoff on the card Pick the payoff on episode 8, or create one

The picker lists every unpaired payoff in the whole book, whichever episode it lives on.

  • Right-click any line in a screenplay → Add note → Add Note / Add Todo / Add Setup (Foreshadow) / Add Payoff (Resolve). The note anchors to that line.
  • Pairing is two-way and crosses files. Both cards then carry an amber chip naming the other, and clicking it opens the other episode and scrolls the card into view.
  • The lightbulb beside the notes panel's title counts paired / total setups across the whole book — the one genuinely season-wide number you get.

The amber bulb in the screenplay's margin does not switch off when you pair a setup. The panel's chip and the lightbulb counter both update correctly; the little bulb in the margin stays amber. Trust the panel.

What is not there: no list of unpaired setups, no per-episode foreshadow report, and no "unpaired" filter. The category tabs are All / Notes / Foreshadow / To-do — payoff notes appear only under All. Pairing is one setup at a time, by hand, and the AI has no tool for it. The full method is in Tracking foreshadowing with notes: setup to payoff.

The beat field inside a scene is a different thing

Board view inside a .script gives every scene a dropdown of fourteen named beats plus a Custom entry — fifteen in the list. It is tempting to assume it feeds the .beats board. It does not, in either direction.

The scene's beat field

  • Fourteen fixed beats plus Custom — fifteen entries
  • A label stored on that scene, in that file
  • Doesn't add up, doesn't warn, doesn't appear on any board

A .beats board

  • Beats you name yourself, from four public structures or none
  • Its own file, book-wide
  • Knows nothing about any scene's beat field

Use whichever you like — just don't wait for one to fill the other in.

When not to do it this way

  • Don't plan the shape on the timeline. It sorts; it doesn't shape. Acts are consecutive, labels are parallel, and mixing them up produces a document that answers neither question.
  • Don't treat the board as a table of contents. A card is a reference to a file, and the board never touches the file tree's order.
  • Don't expect a beat to know it has been written. Nothing marks itself done.
  • Don't lay a season out on a phone. You cannot put a file on a board there.
  • Don't count on a beat board in a delivery. A .beats file has no export button and never appears in a merged PDF, Word, ePub or a beta reader's report. It is a planning surface, not a document. A timeline does export, as a .timeline file for yourself.

Related

Try it in Slima

Open the app and do this with your own book. Free to start, no credit card.

Open Slima
Was this helpful?