Roadmap — what's coming to Clip

Everything in the rest of this guide exists today. This page is the other half of the story: the features we're building next, in enough detail that you can plan around them.

A few honest ground rules before the list:

• Order, not dates. Features are grouped by how far along they are. "Rolling out" means weeks; "In development" means it's being built now; "Planned" means it's designed and queued; "Exploring" means we're committed to the problem and still proving the approach.
• Features ship in slices. Most of these arrive as a first useful version and then grow. The sub-features listed under each item describe where the feature is going, not necessarily what day one looks like.
• The credit rule doesn't change. Anything that runs AI will cost credits, and the cost will be on the button before you press it — same as today. Anything local stays free.
• Some features are plan-gated. Where we already know a feature lands on a specific plan (collaboration on Team plans, for example), it's noted. Everything else follows the usual pattern: core on every paid plan, scale and admin controls higher up.

At a glance

Canvas collaboration

Team plans · In development

Today, a workspace shares projects but a canvas is single-player: teammates take turns. Collaboration makes the canvas itself multiplayer.

What's coming:

• Live presence. See who's on the canvas with you — named cursors, live selections, and a follow mode that puts you in a teammate's viewport while they walk you through something.
• Comments and notes. Pin a threaded comment to any point on the canvas or to a specific asset. @mention a teammate to notify them, reply in-thread, and resolve when it's handled. Notes stay anchored to the thing they're about, even as the canvas grows.
• Reviews. Ask for a review on a frame or a set of assets. Reviewers see exactly what's in scope, leave feedback in place, and hand it back.
• Approvals. A formal sign-off flow for teams that need one: request approval → approver approves or requests changes → the asset carries a visible status badge. An approval history lives with the project, so "who signed off on this" always has an answer.
• Roles that carry over. The organisation and workspace roles you already set (see Organizations & teams) extend onto the live canvas — viewers and commenters can participate in reviews without being able to move a pixel.
• Shared context for the agent. The design chat stays aware of who asked for what, so agent-generated work lands attributed and reviewable like everything else.

Version history already exists per project (see Version history); collaboration builds on it so restores and reviews tell one consistent story.

Avatars

In development

Generate a character once. Use them everywhere, and they stay them.

Today, if you generate a mascot or a model and then generate a second scene, you get a stranger who looks vaguely similar. Avatars fix that: a character becomes a saved, named thing you can drop into any generation.

What's coming:

• Create an avatar two ways. Describe one from scratch ("a cheerful red panda in a courier uniform"), or build one from photos you upload. Either way, Clip distils the character into a reusable identity.
• An avatar library. Your characters live in a library alongside your brand kits — named, previewed, shareable with your workspace.
• Use avatars in any generation. Reference an avatar in an image generation, a video generation, or from the design chat, and the character's face, build, and look carry through the new scene, pose, and lighting.
• A consistency dial. Choose how much locks: face only (change the outfit and hair freely) up to full lock (face, hair, wardrobe, accessories).
• Multi-character scenes. Put two or more saved avatars into the same generation — a recurring cast for storyboards, comics, and campaign series.
• Turnaround sheets. Generate a multi-angle pose sheet for any avatar — front, profile, back, three-quarter — the same way Multi-Angles works on images today (see Reshaping images).
• Not just people. Mascots, creatures, and stylised characters work the same way.

The obvious pairings come with it: brand mascots that stay on-model across a whole campaign, consistent "faces" for social series, and — once Studios arrive — UGC actors and fashion models built on the same foundation. Avatar slots will be capped per plan, and creating avatars from photos of a real person will require that person's consent.

Worlds

In development

A new generation type alongside image, video, and 3D: generate an entire environment — and then work inside it.

A generated image gives you one angle of one moment. A world gives you the place itself: an explorable 3D environment you can move through, stage, and shoot from any direction.

What's coming:

• Generate a world from a prompt or an image. "A sunlit Scandinavian living room", "a neon-soaked market alley at night", or a photo of your actual set — Clip builds the environment around it.
• Explore it. Walk and orbit through the generated space in a viewer, straight from the canvas — the same way 3D models open in a viewer today (see Creating 3D models).
• Stage it. Place your products, avatars, and assets into the world, with lighting and shadows matched to the environment automatically.
• Take renders. Find the angle you want and capture a still — it lands on the canvas as a normal image, fully editable with the whole AI toolbar. Every product shot from every angle, without regenerating the scene.
• Shoot flythroughs. Set a camera path through the world and export it as a video — smooth, deliberate camera moves that prompt-to-video can't reliably give you.
• Edit and extend. Change a region of the world with a prompt ("make the sofa green", "add rain outside the window") and expand the environment beyond its original bounds when you need more room.
• Reuse it. A world is a project asset. Shoot this week's campaign and next week's in the same location.

Worlds are the long way of saying: sets you never have to build, in a studio you never have to rent. Exports will ladder up by plan — stills first, flythrough video next, and 3D exports of the environment itself at the top.

Apps

Rolling out

Clip's canvas and agent can already do almost anything — if you know what to ask for. Apps are the opposite of that: named, one-click workflows for specific jobs, each with a simple form in and a finished, on-canvas, still-editable result out.

The first wave targets the jobs people do most:

• Product Photoshoot — drop in a product photo, get studio and lifestyle shots: clean backgrounds, styled scenes, true-to-product lighting and shadows.
• Mockup Generator — put your design onto real things: devices, apparel, packaging, signage, print.
• 360° Product Video — a smooth spin video of your product from stills. Zoom-and-detail videos for product pages come with it.
• Ad Creative — product plus offer in, a set of ready ad visuals out — automatically resized for every placement (feed, story, banner) with copy that fits each size.
• Social Posts & Carousels — a topic, a URL, or a doc in; a designed multi-slide carousel or post set out, on-brand via your brand kit.
• Thumbnail Maker — title and frame in, a set of high-contrast, click-worthy thumbnail options out.
• Logo & Identity — generate logo directions, then expand the one you pick into a starter identity: colours, type, and a filled brand kit.
• Packaging Concepts — box, label, pouch, and bottle concepts, visualised as 3D-real mockups.
• Marketplace-Ready Photos — one click to platform compliance: pure-white background, correct fill, size, and format for Amazon, Shopify, and friends.
• Variant Maker — one product, every colourway; one design, every language or offer.

Every app puts its result on the canvas as normal, editable assets — an app is a starting point, never a dead end. More apps follow based on what gets used; this catalogue is meant to grow continuously, and Studios will later bundle the right apps for each industry.

Tools

Rolling out

Tools are the single-purpose AI utilities — select a thing, press a button, get the improved thing. Several already live in the image toolbar today (Editing images): background removal, upscaling, eraser, expand, quick edits, text editing, vectorize. The Tools rollout finishes the set and then makes all of it work in batches.

Completing the utility belt:

• Relight — change the light in a shot after the fact: move the key light, warm it up, match a subject to a new background.
• Edit / remove subject — swap, restyle, or cleanly remove the main subject, not just small objects.
• Style transfer — restyle an image or a whole set to a consistent look.
• Video background removal — clean subject cutouts from video, no green screen.
• Auto-captions — burned-in, styled captions for video, in multiple languages.
• Colorize & restore — bring old or damaged photos back.

Batch — the bigger half:

• Apply-to-many. Configure an edit once — remove background, relight, upscale, resize — and apply it to dozens or hundreds of images in one run. Preview on one, commit to all, override individual results after.
• Data-driven batches. Feed a spreadsheet: one row per product, columns mapped to a template's slots (image, name, price, offer). Get one finished design per row.
• Consistency across the set. Batch results are normalised so lighting, framing, and colour hold steady across the whole catalog — the difference between 200 images and 200 matching images.
• Brand rules as a pipeline. Store your rules once — background, padding, logo placement, output formats — and every batch applies and checks them automatically.
• Named outputs. SKU and filename mapping in, correctly named files out, in every format each channel needs.

Batch processing will be a paid-plan feature, with per-image credit costs shown up front like everything else.

Clip MCP

Planned

If you're building your own AI agents, Clip MCP puts Clip inside them. It's an MCP server — the open standard that agents like Claude use to reach external tools — exposing Clip's capabilities to any agent you run.

What's coming:

• Generate from your agent. Images, video, 3D, and audio — your agent calls Clip, the result lands in your Clip project, editable on the canvas like anything else.
• Edit operations too. Background removal, upscaling, quick edits — the toolbar, callable.
• Project and asset access. Your agent can search your projects and assets, pull what it needs, and export finished files.
• Brand-aware by default. Generations dispatched through MCP can carry your brand kit, so agent-produced work is on-brand without your agent knowing what a brand kit is.
• One-click, secure setup. Connect with your Clip account — no API keys to copy around. Your agent gets exactly the permissions your Clip role has, and nothing more.
• Your plan, your credits. MCP usage draws from the same credit balance and respects the same plan entitlements as the app. No separate billing to manage.

The short version: everything the design chat can do for you inside Clip, your own agents will be able to do from outside it.

Connectors

Planned

Connectors link Clip to the services where your assets, your catalog, and your campaigns already live. Three directions, all first-class:

Assets in.

• Google Drive & Dropbox — browse and import from shared team folders without the download-reupload dance; keep a team asset library synced.
• Shopify & product catalogs — your products, names, and prices flow in, ready to drive apps and batch templates.

Context in.

• CRM (HubSpot and friends) — campaign context reaches the design chat, and finished designs are shareable back where the campaign is actually run.
• Calendar — the agent knows the launch is Thursday.

Published out.

• Direct social publishing — schedule and publish finished designs to your social accounts from Clip, sized per platform.
• Shopify listings — push finished product imagery straight onto the listing.
• Slack — generation-complete, review, and approval notifications where your team already talks.

Connectors will roll out a few services at a time, starting with storage and Shopify. If there's a service your team can't work without, tell us — the order of this list is genuinely up for grabs.

Skills

Planned

The design chat is a generalist today — good at everything, specialised in nothing. Skills change that: installable expertise that makes the agent world-class at one specific job.

A skill packages what a specialist knows — the workflow, the checklists, the layout rules, the formats, the taste — so that when you ask for that kind of work, the agent executes like someone who does it every day.

The first skills we're building:

• Poster Design — hierarchy, focal points, print-ready output.
• Reels & Short Video — hooks in the first second, pacing, captions, vertical-first framing.
• Carousel Design — narrative across slides, swipe-through rhythm, consistent system.
• Brand Campaign — one idea, coherently executed across every format at once.
• Packaging — dielines, shelf impact, material-real mockups.
• Thumbnails — faces, contrast, and text that survive being 120 pixels wide.

How they'll work:

• Browse and enable. Skills live in a directory. Turn on the ones that match your work.
• No mode-switching. You don't invoke a skill — you just ask, and the agent brings the right skill to the job automatically.
• Stack with your setup. Skills compose with your brand kit, your styles, and your avatars — expertise on top of your identity, not instead of it.

The library grows over time, and the Studios below each ship with their vertical's skills built in.

Studios

Exploring

Studios are where the roadmap converges: curated workspaces that bundle the right apps, tools, and skills for one industry, so you're never assembling a workflow from parts. Three are planned.

Fashion Studio

For brands, boutiques, and marketplaces that live and die by product imagery.

• Flat-lay to on-model — photograph the garment on a table or a hanger; get it on a model.
• AI models — choose from a diverse model roster, or create a consistent brand model (built on Avatars) who wears every drop, season after season.
• Garment fidelity — patterns, logos, stitching, and drape preserved. The product in the picture is your product.
• Ghost mannequin — clean hollow-form shots for product pages.
• Virtual try-on — show one garment on different bodies and poses.
• Fashion video — on-model motion clips without a shoot.
• Catalog scale — the whole drop through batch, sized for your product pages, synced to Shopify.

UGC Studio

For teams that need a steady stream of authentic-feeling, creator-style content.

• AI actors — a diverse cast of realistic presenters across ages, settings, and shooting styles (selfie, hand-held, product-in-hand).
• Your own presenter — a consistent custom face for your brand, consent-backed, built on Avatars.
• Script to screen — write the script (or have the agent write it — hooks included), pick the actor, get the talking-head video.
• Sound and polish — natural voices, auto-captions, music, and vertical-first framing for TikTok, Reels, and Shorts.
• Localisation — the same ad, natively delivered in other languages.
• Volume — batches of variants: different actors, different hooks, same offer.

Advertising Studio

For performance marketers who treat creative as a numbers game.

• Brief or URL in, campaign out — point Clip at a product page; get ad-ready visuals, copy, and video concepts.
• Variants at scale — hooks, layouts, offers, and formats multiplied into a testing matrix, every placement size covered.
• Hook library — proven opening frameworks for static and video, applied to your product.
• Creative intelligence — guidance on which of your variants is likely to perform, and what to fix, before you spend.
• Straight to channel — publish finished creatives toward your ad platforms via Connectors.

Each studio ships as a curated selection first — the right tools in the right order — and deepens from there. If your industry isn't one of these three, the underlying apps, tools, and skills are still yours to combine; studios are a head start, not a wall.

Have a say

This roadmap moves in whatever order helps the most people. If one of these would change your week — or something you need isn't here at all — tell us through the feedback option in your account menu. "We built it because users asked" is the origin story of half of this page.