Lay your own photographs out on an infinite canvas and arrange them into a sequence. A chat docked to the board reads how they sit, where the hole is, and tells you what is missing using your own archive.
Shooting commercial work? Try the commercial demo →
No credit card required · The demo needs no account

A storyboard made of invented images is a mockup. A storyboard made of your own photographs is an edit — the thing you can actually print, hang, publish or send.
On a board full of pictures, "does this one fit here?" is ambiguous — unless the assistant can see what you marked. Three nested levels turn a gesture on the canvas into the subject of your question.
Every photo you have placed. The widest context — the overall sense of the sequence you are building, and the pool the assistant can profile as a body of work.
Nothing to do. It is what the chat sees by default.
The stretch of the board you are working on right now. Photos under the halo — an ambient circle that follows your viewport — or a set you pick by hand.
Move the halo over the photos, click them one by one, or Ctrl / ⌘ and drag a box.
The individual frames the question is really about. They can sit inside the focus, or be held up against it from elsewhere on the board.
With a focus set, press the spotlight button on any photo.

The finest level you marked is the subject of the question. Everything wider than it is context — which is exactly how you already think when you stand in front of a wall of prints.
Five steps from a pile of frames to an ordered sequence you can defend.
Pull frames from your catalog onto a canvas with no edges and arrange them by hand — left to right, top to bottom, the way the story should read. Nothing snaps to a template; the layout is the edit.
Slide the halo over a stretch of the sequence, or drag a box around it. Spotlight a single frame when the question is about that shot alone. The marks on the board are how you address the assistant.
The arrangement travels with your question as a grid — neighbours, rows, gaps, whether two clusters read as separate groups. So 'does this belong here', 'what breaks the rhythm' and 'what goes in the hole' are answerable, with the position named.
The assistant searches your own library for work that would sit well beside what you have out — matched on look, narrative, palette or setting, your choice. Photos already on the board are held out, so every result is genuinely new. Drag any of them from the reply onto the canvas.
When a reply groups photos into a sequence, one click places them on the board in that exact order, as a block. Then you do what you were always going to do: move two of them and change your mind about the ending.
Before a single word is typed, the board is already a proper editing table: infinite space, real drag-and-drop, and your archive one gesture away from any photo on it.
Pan and zoom across a canvas with no edges. Photos go anywhere, at any size, in any order — no grid, no template, no slots to fill. Dot grid, background colour and auto-align are yours to switch on or off.
Select any photo on the board and a strip fills underneath with visually and narratively related work from your archive. Drag them straight up onto the canvas — the board grows out of the frame you are looking at.
Create, name, switch and duplicate canvases. Each is saved with its own photos, its own arrangement and its own conversation, so a book edit and a client moodboard never collide.
Stack, overlap, cluster, leave a hole where the missing frame goes. Two photos side by side with a third hanging below them means something — and the assistant reads that geometry the same way you do.
Drag a photo to the bin to take it out of the edit. It leaves the board and stops being proposed, but nothing is deleted — reopen the bin and pull it back whenever the sequence changes its mind.
Frame the whole board in one click, spread a pile of overlapping frames apart, or zoom into a run of three. Small tools, but they are the difference between a working table and a mess.
Select photos on the board and push them into a collection, a series, a portfolio, another tool, or a chat — or download them. The canvas is a stage in a workflow, not a dead end.
Open a canvas that already has a real catalog behind it and try the whole thing — arranging, the related-photos strip, the chat — before you commit an archive of your own to it.
"Similar" is not one thing. A frame can rhyme with another by colour and clash with it narratively. Choose which kind of resemblance you are extending, and the strip under the canvas fills with candidates you drag straight up onto the board.
Broad visual and semantic resemblance — scene, subject and overall style at once. The honest default when you just want more of this.
What is happening, and how it feels to watch it. The one that matters when you are building a story rather than a wall of matching pictures.
Palette and colour distribution. For a sequence that has to hold together on a page, or a spread that must not clash.
Places, environments, symbols and cues. Finds the frames shot in the same world, even when they look nothing alike.
Framing, light and treatment. For matching the physical craft of a shot rather than what it depicts.
Match on the concepts the AI already extracted from your archive — pick the ones that matter and search on those alone.
A general photo chat knows your catalog. This one also knows the arrangement in front of you — and that turns questions about sequence, rhythm and gaps into questions it can actually answer.
The board's geometry travels as a grid: what is next to what, what sits above or below, where the empty slot is, whether two clusters read apart. That is why it can say 'it breaks the run in the second row' instead of a generic opinion.
Board, focus, spotlight. You mark the level on the canvas and the finest one becomes the subject of the question — so 'these', 'here' and 'this one' all resolve to the photos you actually meant.
'More like these' anchors on the photos you marked. 'The same palette as these', 'narratively like these', 'in this vein but only verticals' — it switches what similarity means, and combines it with hard constraints when you set one.
Every photo in a reply can be dragged straight from the panel onto the board. Frames already placed are excluded from search results, so nothing it shows is something you have already used.
When a reply proposes a run — an opener, a middle, a closer — one click places those photos on the canvas in that order, as a block. The proposal becomes something you can look at, not just read.
Recurring subjects, dominant lighting, palette, score averages, where and when it was shot. Ask what the canvas is actually about and it answers from the board itself — never confusing it with your catalog at large.
One switch confines every search to the photos already on the board. The conversation stops proposing new material and starts reading what you have: which frames are weakest, what repeats, whether the set holds.
Each canvas keeps its own thread, stored with the board rather than in a chat history. Reopen a project on another device six weeks later and the discussion about that sequence is still there.
The same question means something different depending on what you marked. Write in whatever language you think in — the assistant answers in it.
Any body of work whose order carries meaning.
Find the opening frame, the turn and the closer, and check the middle is not sagging.
Lay out spreads as they will be read, and test whether facing pages actually talk to each other.
Arrange in rows and clusters exactly as they will hang, gaps and all, before anything is printed.
Cut a large body of work down to the frames that earn their place, and defend the order.
Build a board from your own back catalog to show a direction, instead of borrowing someone else's references.
Spread a day's take out, group what belongs together, and find out whether there is a series in there at all.
Every photo it offers has already been searched, scored and understood by the same engines that power the rest of the app.
What photographers ask before storyboarding their own photos with AI
Arrange, question, replace, reorder. The archive is yours and so is the edit — the assistant just never lets you forget what you already shot.
Shooting commercial work? Try the commercial demo — the same board on a product and packshot catalog.