Graphics Library keeps your logos, photos, and design files, labeled by client, project, category, and tags, so anyone on your team finds the right file in seconds. You can drop in a whole folder at once, label hundreds of files in one step, and take suggestions from folder names and from the pictures themselves. Nothing is labeled without your click.
Open Graphics Library →Graphics sits in the sidebar, below Clients. It is also in the Tool library as Graphics Library, and every project page has a Graphics tab that shows that project's graphics.
Once any of your clients has graphics, a Showing row sits at the very top: All (every graphic), then a chip for each client that has graphics with how many it has, then No client. Click one and the library shows just that company's graphics, right there; the chip stays dark so you always see which one you're on, and All goes back. When that client has projects with graphics, a second, smaller Project row lists them (All of that client, then each project with its count). The choice is the page address's ?client= (and ?project=), the same one the sidebar's Working on switch uses, so the two always agree. With a single client the row simply reads All, that client, and No client. On a phone the row scrolls sideways.
Across the top is a strip of counts that are also doors: how many graphics you have (click it to show all), suggestions to review (opens the review queue), your favorites, how many are missing alt text, how many are hidden, and how many are in the Trash. Click one to see exactly those graphics; click it again, or Show all, to go back. A count of zero is shown in gray. Duplicates, next to History and Upload at the top right, opens the duplicates view (on a phone it is under Filters, Needs, Duplicates). An empty library shows only a short welcome with the kinds of files you can bring in and Upload graphics; the counts and filters appear once something is uploaded. When nothing matches a search or filter, the page says so and offers Show all graphics (and Search all graphics when a search and filters are both on).
History, next to Upload, lists every change your team made in the library, newest first, with Undo on each one. New with AI, beside Upload, makes a new picture from your words (or open it straight from /home/graphics?new=ai). Describe it, pick Square, Wide or Tall, tick Words in the picture if it needs lettering (Ideogram spells best), and pick the picture model: ChatGPT Image (best all-round, the default), ChatGPT Image Fast, Nano Banana (Google, great at keeping a face or product the same), Imagen (Google), Ideogram (best lettering) or Recraft (illustration and vector). Each choice shows its tokens per picture, models that aren't connected are grayed out, and your choice is remembered in this browser. Make it shows the picture; Make another tries again; Save to library adds it to the folder Made with AI, marked Made with AI in the Source filter. Nothing is charged when a picture fails.
Press Upload, or just drag files or folders anywhere onto the page. You can also paste a copied picture with ⌘V on a Mac or Ctrl+V on Windows. Choose files and Choose a folder open your computer's file picker. A dropped folder keeps its folder names, including folders inside folders.
Accepted files: JPG, PNG, WebP, GIF, HEIC, AVIF, BMP, TIFF, SVG, AI, EPS, PSD, PDF, and MP4, MOV, M4V, or WebM videos, up to 25 MB each, plus MP3, M4A, and WAV music up to 15 MB each. Each team can keep up to 2 GB; files in the Trash count until they are deleted for good. For a video, your browser grabs a still frame as its poster, so it shows in the grid like any picture, with a play badge and its length; its details play the clip. A clip your browser doesn't play (some iPhone HEVC clips in Chrome) uploads without a poster. Music gets a waveform picture as its preview and shows in the grid with a purple Music badge and its length; its details have a player. aiKit checks that a music file really is MP3, M4A, or WAV when it finishes uploading. Music is listed only in the library itself and in Motion Maker's Music tab: the photo pickers in other tools never show it. Use only music you have the rights to. AVI and MKV files, and converting clips to a web-friendly format, are Coming soon; those files are listed as Not added with that reason.
Each file is read in your browser first: aiKit makes a small preview and a larger viewing copy, and takes two fingerprints, one of the exact file and one of the picture itself. A PDF's preview is a real picture of its first page, and aiKit notes its page size and page count (for example 5.5 × 8.5 in · 2 pages); Illustrator files saved with PDF content get the same. EPS files show the small preview they carry, if any; drawing EPS files that have none is Coming soon. Videos get a poster frame from the clip. Three files upload at a time. The queue shows each file with its preview, folder, size, and what is happening: Reading, Checking for duplicates, Waiting, a progress bar while uploading, then Uploaded. The bar at the top counts how many are done.
If a file stops (a dropped connection, for example), it shows Stopped with a Retry button, and Retry all retries every stopped file. A retry sends only the parts that didn't arrive. If you close the tab in the middle of a big upload, drop the same files again within two days and the unfinished ones pick up where they stopped.
You can hide the upload window while it works; uploads keep going as long as the page stays open. When everything is done, Review this upload shows just those files, ready to label, and Get suggestions for these asks for suggestions on all of them.
Before a file uploads, aiKit compares it with every graphic you can see and with the other files in the same drop. It compares the pictures themselves, not file names, so the same photo saved under a new name, resized, or exported again is still caught.
A match stops that one file and marks it in yellow: Same file as, or Looks like, followed by the matching graphic's name and preview. Skip leaves it out; Upload anyway keeps both. For a big drop, Skip exact copies skips every file that is byte-for-byte the same as one you have, and Upload all look-alikes sends the rest.
Steps at the top of the upload window show where you are: Add files, Label them, and Review. Apply to everything in this upload sets a client, a project, categories, tags, and a rating (Approved or Best, next to the heading) on every file in the drop. You can set them before, during, or after the files upload: as long as the upload window shows them, a change also updates files that already went up, and more files you add in the same window join the same upload. Closing the window, or Clear finished, starts a new upload next time. Same as last batch fills in what you used last time.
Client and Project each end with + New client… or + New project…: type a name and press Add client or Add project, and it is created and chosen right there. A new project is saved on hold, with nothing built; build it later from Projects. A new project is placed under the first client you chose for the upload.
Tags and categories: type a word and press Enter, type a comma, press the Add button that appears in the box, or just click somewhere else. Typed words are never dropped.
When your folders are named after a client or project you have, or carry useful words, From your folder names offers them as chips, for example a folder called "Maple Street Bakery/Menu photos" suggests the client Maple Street Bakery and the tag menu. Click a chip to use it.
Search matches file names, folder names, tags, categories, alt text, and descriptions; every word you type must match. Next to it are four filters for how a graphic is used: Type (Logo, Web graphic, Social post, Print file, Photo, Icon, Illustration, Document or PDF, Music, or No type yet), Client, Project, and Where used (Used somewhere, or Not used anywhere). Each opens a short list with how many graphics each choice has, and a search box when the list is long. More filters (it shows how many of its filters are on) adds Color (swatches for each main color in your library, such as mostly blue or mostly orange), Category (your other categories, such as Products or Places), Tag, Collection (once your team has one), File type (PNG, JPG, SVG, PDF, and every other kind in your library, with counts), Source (Uploaded, Made with AI such as Vectorize Kit logos, or Made in an aiKit tool such as Mockup Studio or Gobo Proof Studio), Shape (landscape, portrait, square), Size in pixels, when it was Added, Added by, Rating, Quality (Top, Good, Low, or not set), and Needs (missing alt text, suggestions waiting, no tags, duplicates, hidden from pickers). Favorites shows only the graphics you starred. Graphics saved before October 5, 2026 show as Uploaded, wherever they came from. Under the filters, your most used tags sit in a row of chips: click one to filter by it, click it again to remove it. Sort (newest, oldest, name, largest files, best quality first, most used first, and Collection order while a collection is chosen) sits at the right, next to your saved views. On a phone, search and a Filters button show first; Filters opens the rest and ends with a Show button that closes the filters and jumps to the results. A saved view's x deletes it.
Filters combine. A bar says how many graphics match and lists each filter as a pill you can click to remove; Clear all removes them all. Every filter lives in the page address, so you can bookmark a view or send it to a teammate, who sees only the graphics they can access. Save this view keeps the current filters under a name you choose; saved views are listed under Saved views above the grid, and pointing at one shows a minus button that deletes it.
Select graphics with the checkbox on a card, or ⌘-click or Ctrl-click a card. Shift-click selects everything between two cards. Once anything is selected, a plain click adds or removes a card. ⌘A or Ctrl+A selects everything that matches the current filters, even graphics that haven't scrolled into view, and Select all matching does the same from the bar. Esc clears the selection.
With the keyboard: the arrow keys move between cards, Space selects, Shift+Space selects a range, Enter opens the card, F stars or unstars it, Delete (or Backspace) asks to delete it, and the Menu key (or Shift+F10) opens its ⋯ menu.
A card's ⋯ menu holds Open details, Send to Motion Maker (photos and videos), Use in a blog post (photos), Download original, Copy link, and Delete. The two Send to choices only appear for tools you can open; see the Kits guide for how sending work between tools works.
The bar at the bottom changes every selected graphic at once: add tags, remove tags, or replace all their tags with new ones; Type sets their type (or Clear the type); add a category; add to a collection (or name a new one); add to a client or a project; set a rating or a quality; Suggest labels; Accept waiting suggestions; and Delete. More holds Add to my favorites, Remove from my favorites, Download as a zip, Remove from the collection you are viewing, and Hide from pickers (or Show in pickers again while viewing hidden graphics). Send to appears when another tool takes everything selected: Send to Motion Maker for up to 40 photos and clips (Motion Maker opens with them on its timeline, in the order you selected them), and Use in a blog post for a single picture (Content Kit opens with it ready to place as a post's main photo). Every change shows an Undo button for a few seconds and stays undoable in History. Graphics you can only view are skipped, and the message says how many. In the Trash the bar offers Restore and Delete forever instead. On a phone the bar stays slim (how many are selected, Clear, and Edit); Edit opens the same choices full width.
Two kinds of suggestions show as purple chips with a plus sign; nothing changes until you click one. From the file name looks at the file and folder names and offers tags, a category, and a client whose name appears in them; it is instant and free. Suggest labels first tries free AI on your own computer (2026-10-13): on a browser with graphics support (recent Chrome, Edge, or Safari on most computers from the last five years) a small open-source model looks at each picture right there, so nothing is charged and only the finished labels are saved; the first time, it sets itself up once (about 150 MB, with a progress line, then it starts at once), and the note says made free on this device. Otherwise, or if it can't start, it sends a small preview of each selected picture (or a video's poster) to aiKit's AI, which suggests a category, tags, alt text, a one-line description, and a quality (Top, Good, or Low), reusing your team's own words where they fit. Cards with suggestions waiting show a small purple dot before their name.
When a graphic has suggestions waiting, its details open with a purple bar, AI suggestions waiting, that jumps to them. Each suggestion is its own row: Category, Tags, Alt text, Description, and Quality. Accept takes that one; Turn down drops it; Edit lets you change it first: for alt text and description it puts the suggested words in the box below, for a category it opens a small box with Save, and for quality it shows Top, Good, and Low to pick from. Each suggested tag has a check (accept), a pencil (edit the word, then Save), and an x (turn down). Accepting tags only ever adds them: your own tags are never removed or replaced. Accept all and Turn down all handle the rest at once. In a graphic's details you can also use the older one-click way: Accept all, or Turn down all. Accept in the bottom bar takes every waiting suggestion on the selected graphics; it fills alt text and descriptions only where they are empty. Files without a preview are skipped. AI suggestions have a monthly limit per team; when it is used up, the message says so, and they start again on the 1st. Name suggestions keep working.
Suggestions to review, in the counts at the top, opens the review queue (its address ends in ?review=1). AI suggestions never change a graphic on their own: they wait here until a person approves them.
Waiting shows one card per graphic, oldest first: the picture, then Categories and Tags with the suggested ones already merged in (the words that are new from the suggestion are listed under the tags), Alt text, Description, and Quality, where the AI's pick is marked suggested. Edit anything, then press Approve to save exactly what the card shows. Turn down keeps the graphic as it was and clears the suggestion. Hide graphic keeps it in the library but out of pickers and suggestions. With a card selected, A approves and R turns down.
Approve these as suggested approves every card on the screen at once; it fills alt text, descriptions, and quality only where they are empty and adds the suggested tags and category. Each decision shows Undo for a few seconds and stays undoable in History.
Reviewed lists graphics approved or turned down lately, with when, so a wrong call is easy to fix: Fix labels opens the graphic's details.
Favorites are personal: the star on a card (or the star at the top of its details, next to the close button) keeps a graphic in your own Favorites, and nobody else's list changes. Favorites in the filters, the favorites count, and the Favorites tab in every picker show them.
Quality (Top, Good, Low) says how usable a graphic is on a professional page. It is separate from Rating (Approved, Best, Not using), which records your decision about it. Suggestions in pickers rank Top first; Best quality first in Sort does the same in the library. A card shows at most one flag in its top corner (Best, Hidden, or its place in a collection), and its line under the name says how many places it is used (Used 2×). The flag steps aside for the select box when you point at the card.
Hide keeps a graphic in your library, and anywhere it is already used, but leaves it out of pickers and suggestions. Hidden ones leave the main grid; the hidden count at the top lists them, with Show in pickers again. If the graphic is used somewhere, aiKit says where before hiding it.
Collections are your team's own named groups, such as Spring promo or Homepage picks, separate from categories. A graphic can be in many. Add to a collection in the bottom bar or in a graphic's details adds to one or makes a new one. Choosing a collection in the filters shows it in its own order, with each card's position, and offers Arrange (drag cards, type a position, or use Move earlier and Move later, then Save order), Rename, and Delete collection, which deletes only the group; its graphics stay. Only people on your team see collections.
Duplicates, at the top right next to History, groups every graphic you can see with its copies: Same file for byte-for-byte copies, Look alike for the same picture resized or saved again. Each card in a group shows its size in pixels and bytes, folder, how many labels it has, and how often it is used, so you can keep the best one.
Keep this one moves the others in that group to the Trash in one step, with Undo; they wait there 30 days like any deleted graphic. Graphics you can only view are left alone.
Click a card to open its details on the right. The address changes to a link for that one graphic, which Copy link copies; only people who can see the graphic can open it. The panel shows a large preview, Download original (the exact file you uploaded), View full size, Copy link, and Delete, then the file and folder names, type, size in pixels and bytes, its Source (Uploaded, Made with AI, or Made in an aiKit tool), and who added it when.
Alt text is the short description websites and screen readers use; under 125 characters reads best, and the counter shows how close you are. Alt text and Description save by themselves a moment after you stop typing; Saving… and Saved show next to the Words heading, and each save can be undone like any other change. Below it are Description, Type (tap Logo, Web graphic, Social post, Print file, Photo, Icon, Illustration, or Document or PDF to set or clear it; a graphic can have more than one, and PDFs always count as documents), Other categories, Tags (type and press Enter; matching words you already use appear as you type), Rating (Not reviewed, Approved, Best, Not using), and Clients and projects, where you see every client and project the graphic belongs to, add more, or remove one. A graphic can belong to many clients and projects. Changes save as you make them.
View full size (or a click on the big preview) opens the viewer: the graphic as large as your screen allows, on a dark background. Zoom with the minus and plus buttons, the + and − keys, or ⌘ or Ctrl with the scroll wheel, from 25% to 800%; Fit (or 0) shows it whole, and Actual size shows a picture pixel for pixel. Drag to move around when zoomed in. Photos load the original file for full detail. A PDF shows every page, drawn sharp at the current zoom, each labeled Page 1 of 2 and so on. Beside the picture are the file's details, Download original, Add to favorites, Tags, Categories, Clients and projects (with + New client and + New project), and Add to a collection, all editable in place. Esc closes it. The viewer has its own link: the graphic's link with &view=1 added.
The details also show Quality (Top, Good, Low, or Not set), and Collections, with links to each collection and a way to add more. Videos play right in the panel, and their type line shows their length.
Focus and crop sets what must stay in view when the graphic is cropped. Click the part that matters, or move the point with the arrow keys (Shift moves it further), and set Zoom. Four previews show the result live as Square (1:1, social posts), Portrait (4:5, feeds), Wide (16:9, banners and video), and Link preview (1.91:1, shared links). Save focus keeps it; Reset to center clears it. Cards in the library and in every picker use the same focus, and kits get it with the graphic.
Enlarge 2× (on pictures smaller than 1200 pixels, like small logos or low-resolution photos) makes a sharper copy twice the width and height, free, on your own computer with a small open-source model (set up once, about 8 MB). It takes about half a minute and saves a new graphic next to the original, with its labels, client, and project, named with enlarged 2x; the original is never changed. It needs a browser with graphics support; on others it runs more slowly.
Looks similar shows up to eight other graphics whose pictures are closest to this one, such as other versions, crops, or shots from the same set. Click one to open it.
Where it's used counts and lists every place the graphic is used: its projects, and any blog post, email, product, or page a kit placed it on, each a link.
History lists every change to this graphic, who made it, and when. Undo puts the earlier values back. If someone changed the same thing again since, their newer value stays and the message says so. Downloads are listed too.
Hide from pickers sits at the bottom. Delete is next to Download original at the top, and in the full-size viewer's top bar.
There are four ways to delete: the ⋯ button at the bottom right of any card (it also has Open details, Download original, and Copy link), Delete in a graphic's details or in the full-size viewer, Delete in the bottom bar for everything selected, and the Delete key on a card you moved to with the arrow keys. Each one opens the same short check before anything happens.
The check looks up where each graphic is used and lists it: web pages a kit placed it on (marked Live on the site's address when the page is published), blog posts, emails, and product pages, the projects it belongs to, and how many Hub items link it. When it isn't used anywhere, it says so. It also says what happens: the graphic moves to the Trash and leaves the library, project lists, and pickers right away, but nothing that uses it breaks. Pages, posts, emails, and proofs it was placed on, and Hub items that link it, keep showing it, and aiKit keeps its file until it's taken off there, even after 30 days. Published sites also keep their own copy. Take it off those places too if you no longer want it shown. Press Delete, or Keep it to change nothing.
Right after you delete, a bar at the bottom says what was deleted, with Undo. Undo, Restore, or History bring a graphic back exactly as it was: its labels, clients, projects, and collections.
The Trash count at the top opens the Trash. A note there says how many graphics it holds and how much storage they take up. Each card shows how many days it has left (or Still in use, kept), and its ⋯ menu offers Restore, Open details, and Delete forever. A graphic stays in the Trash 30 days after it was deleted, then a daily check deletes it for good, along with its file. Delete forever (one graphic, or several selected) and Empty the Trash do that right away and say how much space they freed; there is no Undo for them. A graphic still used on a page, post, email, or proof, or linked or mentioned in a Hub item (even one in the Hub's trash), is never deleted for good, by the daily check or by you, until it's taken off there; until then it stays in the Trash and keeps working where it's used, and Delete forever says so. Files in the Trash count toward your 2 GB until they are deleted for good.
Only people who can change a graphic can delete or restore it: your team, and staff on one of its clients or projects. Reviewers and clients never see Delete. Delete forever and Empty the Trash are for the team that uploaded the graphics.
A project page's Graphics tab shows that project's graphics. Add from library opens the shared picker: Suggested ranks your graphics for this project first, then All, Favorites, and Collections; pick as many as you like and press Add to this project. For a client's project, the picker shows that client's graphics and general ones, never another client's. Hidden graphics and ones rated Not using stay out. Upload opens the upload window with the project already chosen. Remove takes a graphic off the project only; it stays in the library and in any other project. Open in library shows the same graphics with every filter.
The picker on a project's Graphics tab is built to be shared: the Blog, Product, and Email kits get the same one (Coming soon). Its Suggested tab ranks your graphics by how well their tags, categories, alt text, description, and file names match the page's words, counting rare words more than common ones and treating forms like photo and photos as the same word. Top quality, Best, graphics already in the project or client, and your favorites move up; graphics used in many places move down, so pages don't all repeat one photo. Each suggestion shows the words it matched.
Graphics already on the page are never suggested again, and neither are their look-alikes, and one list never holds two look-alikes. Type extra words in the search box to steer it. The picker remembers its last tab and search for each project. Once the kits use it, every graphic they place is recorded in Where it's used.
Have a quicker, specific question instead? Check the FAQ.