Updated November 10
A practical field guide to AI for working designers: what changed, why it matters, what's worth trying, and what could go wrong.
98 articles
3 min read· 3 min read
Announced at Adobe MAX, Firefly Design is a new beta workspace inside Firefly that brings illustration, image editing, animation, 3D, and print onto one canvas to create, review, and deliver. Adobe has shared the pitch, not the details, so treat it as something to explore rather than build a job on.
Brand & Graphic
3 min read· 3 min read
Generative Restyle is out of beta. You describe a new look for a photo, and Photoshop puts it on its own editable layer with a slider for how strong it is, so you can find the right amount without rewriting the prompt.
Photo & Imaging
3 min read· 3 min read
Google's video editor moved to its newest video model, with longer clips, a duration slider, and steadier characters and products across shots. Any Google account gets a small free monthly allowance, but Google's own pages don't agree on exactly how small.
Motion & Video
6 min read· 6 min read
One click gets you most of a portrait cutout. The hair is where it falls apart. Pulling together Adobe's documentation and several independent tutorials, here's how to choose between Cloud and Device, refine the edge without turning hair into a helmet, and deal with the color fringe that's usually left behind.
Photo & Imaging
4 min read· 4 min read
Viggle's new video-to-video tool lets you chat your way to a different actor, outfit, background, or weather in an existing clip, and the model weights are out too. The app version is capped at 720p, and the open version needs serious hardware and a careful read of its license.
Motion & Video
4 min read· 4 min read
Kapwing now has an MCP server that hooks its video tools into Claude and ChatGPT, so you can upload footage, add subtitles, translate, dub, and pull an export from a chat. Its own help pages don't fully agree on what it can do yet, or on who gets it.
Motion & Video
5 min read· 5 min read
Since September, thread after thread on both forums has been about AI entry points that pop up uninvited and are hard to dismiss. The fixes exist, but they're buried, and some people want a lot more than a hidden setting.
UI/UX, Photo & Imaging
4 min read· 4 min read
The update adds animated GIFs in ChatGPT and the API, brings photo markup to the API, and adds SVG output for developers. The SVG side is the more interesting one for designers, but it's API-only for now and comes with real limits.
Brand & Graphic, Motion & Video
3 min read· 3 min read
Agent-made vectors now come with sensibly grouped layers and real strokes for line work, and you can attach your own illustrations as a style reference. Here's what that changes, and what Figma hasn't said yet.
Illustration, UI/UX
3 min read· 3 min read
ON1's new release groups similar frames and flags closed eyes and soft focus, alongside a rebuilt RAW engine and new NoNoise AI models. The culling help is a MAX-only feature, and ON1 is light on detail about how it works.
Photo & Imaging
3 min read· 3 min read
The unified Affinity app is now on iPad with Pixel, Vector, Layout and a new Sketch Studio, it opens the same .af files as desktop, and only the Canva AI features cost anything.
Brand & Graphic, Illustration, Photo & Imaging
3 min read· 3 min read
Sparseal's retopology, UV and baking app is now on desktop. You sketch guides and the AI fills in the quads, and the models run on your machine, trained only on synthetic data.
3D
4 min read· 4 min read
Framer's Convert add-on now works with the Framer Agent, so you can ask it to build, edit, and analyze A/B tests, funnels, and triggers. Convert also picked up country-based triggers, and a rebuilt Agent harness gives outside agents the same reach through MCP.
Web, UI/UX
6 min read· 6 min read
Lightroom and Photoshop now give you several ways to enlarge a photo, from free Super Resolution to credit-based Topaz Gigapixel and Bloom models. Pulling together Adobe's documentation and several independent tests, here's how to choose one by subject, prep the file, and check the result where it counts: on paper.
Photo & Imaging
4 min read· 4 min read
Pixelmator Pro gets a describe-what-you-want fill, Final Cut Pro absorbs MotionVFX, and Keynote, Pages, and Numbers open up to outside agents through MCP. It all lands in November, and Apple left a few big questions unanswered.
Photo & Imaging, Motion & Video
5 min read· 5 min read
Most people use Boards as a place to dump generations. Across Adobe's tutorials and independent write-ups, a better pattern shows up: boards with one job each, references that steer generation, and a clean handoff to Photoshop.
Brand & Graphic, Photo & Imaging
4 min read· 4 min read
Research docs, spreadsheets, and images can now feed straight into a Figma prompt, even on Starter. It's handy, but the setting that decides what the agent can do on its own deserves a minute of your time first.
UI/UX
4 min read· 4 min read
Paste a link and Krea builds a kit with your logo, fonts, selling points, and products, then generates ads from it. The useful part is the step in between, where you get to fix what it found before anything gets made.
Brand & Graphic
3 min read· 3 min read
A new Griptape node connects Foundry's compositor to its AI orchestration platform, so prompts, models, and seeds sit right in Nuke. It's a beta, the AI usage costs extra, and where your plates go depends on how you run it.
Motion & Video
3 min read· 3 min read
A few minutes of walking around with a phone can become a 3D scene at real-world scale, with the gaps filled in. Before you scan a client's space, check the plan you're on and what the terms let SpAItial do with your upload.
3D
4 min read· 4 min read
A week after Craft, Canva has a second in-house model, and this one works on designs you already have. It's in Canva AI for Pro subscribers now, and the speed and cost numbers are Canva's own comparison against its old agent.
Brand & Graphic
5 min read· 5 min read
Describe an app in plain language and Runway builds and hosts it, wired to your models and Brand Kit. It's on every paid plan, but check the minimum credit balance and the conflicting word on public links first.
Motion & Video, Brand & Graphic
7 min read· 7 min read
Object Matte replaced Roto Brush in After Effects 26.2, and it changes the job from painting a selection to clicking on one. Pulling together Adobe's guidance, early coverage, and what users are running into, here's a workflow from first click to a matte you can actually composite with.
Motion & Video
3 min read· 3 min read
KeenTools, the team behind FaceBuilder, now has a browser Studio and an API that reconstruct a textured head with consistent topology and ARKit blendshapes from your photos. It's pay-per-head, and the fine print is worth reading before you build a pipeline on it.
3D
7 min read· 7 min read
Webflow can draft alt text for a whole site in one pass, and Framer plugins can too. Accessibility guides agree the drafts describe what an image looks like, not why it's on the page. Here's how to sort, generate, and review so the result actually helps screen reader users.
Web
3 min read· 3 min read
A new importer turns old .fla and XFL files into Rive projects: symbols become artboards, frame labels become state machine states, and buttons, tweens, and sounds carry over.
UI/UX, Motion & Video
4 min read· 4 min read
Framer's official MCP server connects outside AI tools straight to your projects. The agent can edit the canvas, manage CMS content, and publish, and nothing needs to be open or installed on your machine.
Web, UI/UX
4 min read· 4 min read
Canva's first in-house image model sketches a written plan of the composition and any text before rendering, which is meant to make its images easier to split into editable layers with Magic Layers.
Brand & Graphic, Photo & Imaging
3 min read· 3 min read
Ideogram's first vector model skips the pixels entirely. It generates SVG artwork from a text prompt, or converts an existing image, and the pitch is shapes and colors you can actually edit afterward.
Brand & Graphic, Illustration
3 min read· 3 min read
Lightroom on mobile now spots closed eyes and offers a one-tap generative fix called Open Eyes. Adobe calls it free, but early hands-on reports say it spends generative credits, and the new eyes aren't always a perfect match.
Photo & Imaging
3 min read· 3 min read
Spline now hosts its MCP server, so you can connect ChatGPT, Claude, Cursor, and other AI tools by signing in with your Spline account, no desktop app or API key required. The same update brings PBR materials and timeline animations to native iOS and Android exports.
3D, Web
6 min read· 6 min read
Cycles' denoiser is on by default, so most people never touch its settings. Pulling together Blender's manual and several long Blender Artists threads, here's how to set it up for stills and animation, and where the advice disagrees.
3D
3 min read· 3 min read
Scene Light, new in Red Giant 2027.0, uses AI to estimate depth and surface direction from your footage, then lets After Effects lights fall across the scene, even behind people and objects in the foreground. Maxon says the analysis runs on your own machine.
Motion & Video
3 min read· 3 min read
The perspective and distortion tool can now select a single subject and reshape just that part of the frame, opens Raw files directly, and DxO is clear the AI only selects, it never fills or invents anything.
Photo & Imaging
3 min read· 3 min read
Open a report, benefits guide, or spec sheet in Acrobat and ask for a flyer, social post, or newsletter. You get up to three designs to refine by chat or finish in Adobe Express.
Brand & Graphic
7 min read· 7 min read
Firefly Custom Models can learn your illustration style from 10 to 30 images. Pulling together Adobe's docs and several independent write-ups, here's how to curate, caption, test, and actually use one.
Illustration, Brand & Graphic
3 min read· 3 min read
Krea's new editing model returns most 2K edits in under 8 seconds, takes several reference images per edit, and is pitched at keeping labels, logos, and faces intact while the scene around them changes.
Photo & Imaging, Brand & Graphic
3 min read· 3 min read
Agents connected to Webflow can now read page-level Analyze traffic, Optimize experiment results, and AEO Analytics, and Google Gemini joins Claude and ChatGPT as a way to connect.
Web
7 min read· 7 min read
Cinema 4D now ships with its own MCP server, so Claude or ChatGPT can rename, reorganize, and check your scenes for you. Pulling together Maxon's docs and several early write-ups, here's how to set it up and where it actually earns its keep.
3D, Motion & Video
6 min read· 6 min read
Turntable can spin a flat character or object around and hand you dozens of editable views, but it's picky about what you feed it. Pulling together Adobe's docs, hands-on reviews, and early tester reports, here's how to prep art for it and what to do with the results.
Illustration, Brand & Graphic
4 min read· 4 min read
At Config India, Figma opened code-backed prototypes on the canvas to everyone and introduced Specs, a Markdown document that lives next to your designs and can be handed to a coding agent. Code layers were free to create and edit on paid plans through November 4, and now use AI credits.
UI/UX, Web
3 min read· 3 min read
Prompt a site in Figma Make and it can now set up a database, file storage, user login, and an admin panel your client can edit, all hosted by Figma. The hosting is free while it's in beta, and Figma says pricing will be announced before it takes effect.
Web
5 min read· 5 min read
Enhance Speech can rescue a phone-mic interview in one click, and it can also leave a voice sounding hollow or robotic. Pulling together Adobe's own notes, several guides, and what editors report in Adobe's forums, here's how to use it without overdoing it.
Motion & Video
3 min read· 3 min read
Google opened up its newest frontier model on October 13, and Figma Make got it the same day. In the Gemini app it's for paying subscribers only, but in Figma Make it's listed for all plans. Figma warns that credit use for non-default models can vary a lot.
UI/UX, Web
6 min read· 6 min read
Auto Style reads a block of text, works out what's a heading, a list, or body copy, and applies a Style Pack of your own paragraph styles. Set it up carefully and it turns "can we see it in a different font?" into a one-click change.
Brand & Graphic
4 min read· 4 min read
Relight, a generative layer that started in the Photoshop beta, is now in the main Photoshop 28.0 release. You drag lights around the canvas, set their color and softness, and Photoshop replaces the photo's lighting on a separate layer. Each final render costs 30 generative credits.
Photo & Imaging
6 min read· 6 min read
Sampler's AI-powered Image to Material can pull a full PBR material out of a single phone photo. Getting one that tiles cleanly and holds up on a render takes a few more steps. Here's the path, pulled together from Adobe's docs and several independent guides.
3D
6 min read· 6 min read
FigJam can group a wall of interview stickies by topic and write a summary of each pile in seconds. Pulling together Figma's docs, a competing approach in Miro, and what researchers have found when they test these tools, here's how to use that speed without letting it make your calls.
UI/UX
3 min read· 3 min read
A new Rerun (Thinking) button on V8.2 jobs has the model look for where it missed your prompt before generating again. The same alpha update adds shared folders, so a team can generate into one place.
Illustration, Photo & Imaging
5 min read· 5 min read
More clients show up with a logo they made in an image generator and a request to "just clean it up." Drawing on several designers' write-ups and general vector practice, here's how to scope the job, rebuild the mark properly, and have the ownership conversation before you deliver.
Brand & Graphic
6 min read· 6 min read
Lightroom doesn't copy an AI mask from one photo to the next. It copies the instructions and redraws the mask on every frame. Pulling together several independent guides and Adobe community threads, here's how to batch subject, sky, and people masks so they actually land where you meant them to.
Photo & Imaging
6 min read· 6 min read
Content-Aware Fill can erase a boom mic, a sign, or a passerby from a shot, but only if you pick the right fill method and feed it the right mask. Pulling together Adobe's docs and several independent tutorials, here's a fuller workflow, including where the guides disagree.
Motion & Video
6 min read· 6 min read
A client sends a flattened JPEG or a file with every word outlined, and nobody knows the typeface. Pulling together Adobe's docs and several independent guides, here's how to get good matches out of Retype, what it no longer does, and how to check a match before you commit to it.
Brand & Graphic
4 min read· 4 min read
Anthropic put Claude Motion into beta for Team and Enterprise plans. It builds short, code-driven animations you can export as MP4 and send on to Adobe and other editors. The same day, Claude Design, Docs, and Slides left beta and came to every plan, Free included.
Motion & Video, Brand & Graphic
4 min read· 4 min read
Black Forest Labs' new image model takes a layout, not just a prompt: you draw a box for each element, describe it, and the model builds the image around your composition. You can then edit, swap, or move one region and tell it to leave the rest alone.
Photo & Imaging, Brand & Graphic
5 min read· 5 min read
Lightroom's Remove tool now packs Generative Remove plus one-click Distraction Removal for people, reflections, and dust. Pulling together Adobe's own guidance and several independent guides, here's how to pick the right mode, brush it properly, and know when to hand off to Photoshop.
Photo & Imaging
6 min read· 6 min read
Illustrator's Mockup feature wraps vector art onto a photo of a real object and keeps it editable. Pulling together Adobe's docs and several independent write-ups, here's how to set it up, which photos it handles well, and where you'll still need to finish by hand.
Brand & Graphic
4 min read· 4 min read
Luma's new Ad Variants resizes and translates a finished static ad into standard placements, and Higgsfield's Ads Studio now builds static ads from your site and remakes video ads with your own product, cast, and script. One week, one tedious job.
Brand & Graphic, Motion & Video
4 min read· 4 min read
As of October 6, the agent in Figma Design is generally available, with library guidelines written in markdown, file search inside chat, and a catch: it no longer runs for free.
UI/UX
6 min read· 6 min read
Harmonize matches a pasted-in subject's light and color to its new scene in one click. Pulling together Adobe's own tutorial and several hands-on guides, here's how to set it up, where it goes soft, and how to finish the job by hand.
Photo & Imaging
3 min read· 3 min read
Muse for Small Business connects Meta's AI agent to Canva, Figma, Shopify, and a long list of other work tools, which means some of your small-business clients may soon be making on-brand graphics by chatting with it.
Brand & Graphic
6 min read· 6 min read
Style references, sref codes, moodboards, and personalization all promise a consistent look, and they behave quite differently. Pulling together Midjourney's docs and several practitioners' write-ups, here's how to build a look that holds across a whole campaign.
Photo & Imaging, Brand & Graphic
3 min read· 3 min read
Ideogram's new model is aimed at the stage after generation: changing one part of an image, over and over, without the rest of it slowly drifting, softening, or shifting color.
Photo & Imaging, Brand & Graphic
6 min read· 6 min read
Generative Extend can save a shot that's a few frames too short, and it can also hand you a brightness bump or a soft, flickery tail. Pulling together Adobe's own notes and several hands-on reviews, here's how to use it where it holds up.
Motion & Video
4 min read· 4 min read
The deal closed September 23 with promises that Topaz stays standalone. Across Topaz's own forum and other creative communities, the reaction is less outrage than guarded skepticism about pricing, licenses, and credits.
Photo & Imaging, Motion & Video
6 min read· 6 min read
Meshy, Tripo, and similar tools get you a convincing model in minutes, with geometry that's rarely ready for anything. Pulling together Blender's manual and several independent guides, here's a cleanup path and where the advice disagrees.
3D
6 min read· 6 min read
Figma can build custom shader fills and effects from a plain-language prompt, including animated and cursor-reactive ones. Pulling together Figma's docs, independent write-ups, and forum threads, here's how to use them and where they still fall short.
UI/UX, Web
3 min read· 3 min read
Webflow added agentic translation to Localize: a one-time pass you review, or a continuous mode that retranslates every change to your primary content, plus rules for keeping some content out of AI's hands entirely.
Web
3 min read· 3 min read
Three months into its open beta, Figma Motion picked up the features that take it from something to experiment with to something a team can standardize on: reusable styles you publish to a library, audio on the timeline, and Lottie output.
UI/UX, Motion & Video
6 min read· 6 min read
Text to Vector Graphic gives you real paths, but often far more of them than you'd draw yourself. Pulling together Adobe's docs and several independent guides, here's a workflow from prompt to a file you'd actually hand off.
Illustration, Brand & Graphic
6 min read· 6 min read
Magic Mask can isolate a moving person or object without rotoscoping, but most frustration comes from over-guiding it and fixing the wrong frame. Pulling together several guides, here's a workflow from first click to a cached matte you can grade against in real time.
Motion & Video, Photo & Imaging
3 min read· 3 min read
Canvas lays out every page of a Shopify store on one zoomable workspace and has Sidekick, Shopify's AI agent, write the theme code as you talk. It's early, and the fine print matters a lot if you build stores for clients.
Web
5 min read· 5 min read
Most people try Generative Recolor once, get a random palette, and move on. Pulling together several tutorials, here's a fuller workflow: prep the art so the AI can actually see your colors, explore broadly, save what works, then snap it back to a real brand palette.
Illustration, Brand & Graphic
3 min read· 3 min read
Kling's new video model is now available worldwide on its website, with single-pass clips up to 30 seconds, up to 15 references, and 10 keyframes. The promised 10-bit HDR and 21:9 output still aren't here, and API access is coming later.
Motion & Video
3 min read· 3 min read
Lightroom desktop's September 29 update adds an early access Prompt to Edit tool powered by Google's Nano Banana model. It's handy for restoration and big changes, but it produces a new generated image, not an edit you can dial back.
Photo & Imaging
3 min read· 3 min read
A mid-September Firefly update adds automatic subject detection to background removal for video, applying it across an entire clip instead of frame by frame.
Motion & Video
3 min read· 3 min read
Paste a company website or upload a brand guidelines PDF, and Canva's Brand Kit Builder pulls out logos, colors, fonts, and brand voice on its own, rolled out as part of Canva's September 2026 feature drop.
Brand & Graphic, Web
4 min read· 4 min read
Developer Vishwas Gopinath's write-up on connecting Claude Code to Figma's MCP server still holds up as a technique, though Figma has since replaced his exact setup steps with a plugin-based flow. Here's why it's worth doing and where Figma's current instructions actually live.
Web, UI/UX
3 min read· 3 min read
Independent essays published months apart keep landing on the same diagnosis: AI design and coding tools converge on the same look, and the fix is deliberate constraint, not a better prompt.
UI/UX, Web
4 min read· 4 min read
Across months of threads on Figma's community forum, the same pattern keeps showing up: Figma Make's credit system charges for the corrections its own AI output makes necessary, and people are running out fast.
UI/UX, Web
4 min read· 4 min read
Designer Roberto Moreno Celta's published breakdown of why most Figma variable setups collapse, and the naming and handoff workflow he uses instead to keep design and code actually in sync.
Web, UI/UX
3 min read· 3 min read
A signed add-on now lets Blender artists generate meshes, rigs, and scene elements from inside the 3D viewport, and get back editable data instead of a locked render. Early coverage is cautiously optimistic, not sold.
3D, Motion & Video
5 min read· 5 min read
Most people who use Firefly inside Photoshop know two moves: fill a resize, erase an object. A look across Adobe's own documentation and independent tutorials turns up a model picker, a free removal path, and a reference-image workflow most people never open.
Photo & Imaging, Brand & Graphic, Web
3 min read· 3 min read
The Adobe for Claude plugin now handles PDFs and lets you step into a layer-based Express editor yourself, while Photoshop, Lightroom, Express, and Firefly tools arrive in Google Gemini.
Brand & Graphic, UI/UX
3 min read· 3 min read
Adobe's new assistant can reorganize messy projects, write expressions from plain-language descriptions, and handle multi-step chores, but for now it only lives in the After Effects beta.
Motion & Video
2 min read· 2 min read
Krea's AI agent, which plans a job, picks from more than 150 models, and generates or edits images, video, and audio from one prompt, no longer needs an invite.
Photo & Imaging, Motion & Video
2 min read· 2 min read
A quiet Firefly update triples the crop options for GPT Image 2 on the web and in Boards. Small on paper, useful in practice if you generate assets for more than one placement.
Photo & Imaging, Illustration, Brand & Graphic, Web
3 min read· 3 min read
Canva has grouped its four pro acquisitions into one connected suite, and the headline Affinity 3.3 update adds a Blend Tool, InDesign file import, JavaScript scripting, and a paid Layer to 3D feature.
Brand & Graphic, Illustration, Photo & Imaging, Motion & Video
3 min read· 3 min read
A write-up of Nolen Jonker's XDA walkthrough: connect Claude to Affinity, let it do the tedious layer renaming and color swaps, then keep the good ones as scripts. Updated for Affinity 3.3's new Scripting Studio.
Brand & Graphic, Illustration
3 min read· 3 min read
CorelDRAW's September 2026 update generates editable vector artwork from prompts or existing art, speeds up bitmap tracing, and lets CorelDRAW Web use your own fonts and export DWG and DXF.
Brand & Graphic, Illustration
3 min read· 3 min read
Claude, Claude Code, and ChatGPT Codex can now organize media, cut a highlight reel, or batch-render inside Resolve through plain language, using a native MCP server built into Resolve Studio. Studio tier only.
Motion & Video
2 min read· 2 min read
A new 'Figma node' lets you paste a Design frame straight into Weave, so it becomes a live, on-brand source that AI workflows can build on at scale.
Brand & Graphic, UI/UX, Motion & Video
3 min read· 3 min read
Framer's Skills feature lets you save reusable instructions, for a design system, a writing style, or a CMS workflow, that its AI agent follows across tasks instead of you re-explaining your conventions every time.
Web, UI/UX
3 min read· 3 min read
A new alpha edit model lets you open any generated image, describe a change in plain language, attach up to four reference images, and see every edit in a session lined up in one place.
Photo & Imaging, Illustration
2 min read· 2 min read
A new fast tier of Recraft's image model trades reference-image input and aspect-ratio range for generation times of about 1.4 seconds, at $0.007 an image.
Illustration, Photo & Imaging, Brand & Graphic
2 min read· 2 min read
AI agents connected through Runway MCP can now find and reference Brand Kit assets, like logos, colors, and fonts, when generating images and video. Creating or editing a kit still has to happen in the product.
Brand & Graphic, Motion & Video
3 min read· 3 min read
Spline's V2 relaunch adds an in-editor AI agent that can build and edit full 3D scenes from a prompt, plus an MCP server so tools like Claude Code and Cursor can drive the editor directly.
3D, Web
2 min read· 2 min read
Agent Instructions, Webflow's context files for AI agents working on a site, can now be generated from a site you've already built, covering brand voice, design system, assets, and CMS content.
Web, UI/UX, Brand & Graphic