Which AI tools do graphic designers actually use, and what for? This page collects real examples: studios, freelancers and in-house teams using tools like Midjourney, ChatGPT, Claude, Adobe Firefly, Runway and Figma Make for branding, illustration, packaging, motion and web work. Each entry says what they did, how it went and where it fell short, with a link to the original source. 4 The Creatives finds and summarizes these accounts; we haven't tested the tools ourselves.
44 real examples and 24 tool profiles · Updated Oct 31, 2026
Logo exploration in Midjourney and Ideogram, then vectorized and redrawn by hand; brand systems, type and brand documents built with Claude; campaign and event visuals.
Noo–Studio's record of a Claude agent drawing a variable icon font in Python while a second, no-edit agent judged it against numeric rules.
Noo–Studio turned the N and O of its logo into a variable font in under three days: AI agents drew and checked glyphs, the founder judged them.
A designer’s fictional fintech brand uses Midjourney and Ideogram for exploration, then redraws the chosen mark point by point in Illustrator.
Claire Vo reports importing a public site's assets to build a design system, then generating a landing page and deck from it.
A seller’s own example: designers refine a generated logo for a fictional wellness brand in three rounds, with a quoted €199 price and 72 hours.
Daria Cupareanu describes building a brand skill in Claude so documents match a brand without manual restyling.
A studio’s published guide covers choosing a traceable concept, hand-tracing, colour specs, a logo family, type licensing and file delivery.
A designer’s 2025 concept exercise combines ChatGPT, Sora and Nano Banana for a studio identity, finished in Illustrator and Photoshop.
In a 2023 X thread, Linus Ekenstam describes a barbershop logo built from Midjourney images, vectorized, then tidied in a vector editor.
A class project chains several AI tools into a brand system and site; the author records what he had to correct by hand at each stage.
An agency says it used ChatGPT with the client’s real product photos as reference, then made static and animated visuals in Canva.
A designer reports extending a small brand from a Canva logo into 51 packaging designs, guidelines and an AI photo workflow the client reuses.
Zapier’s brand team describes building an event identity with Midjourney imagery, automated session assets and personalized badges.
Prompt methods for consistent images, icon sets at scale, color variations and packaging exploration with Firefly.
A design news report says Pentagram trained an illustration style from paint and cut paper, then produced reusable icons for federal agencies.
Adobe’s 2024 account says Mattel explored packaging concepts with generated imagery, conventional editing and 3D presentation.
Adobe’s 2023 profile covers an artist using generative recolouring alongside her hand-lettering and illustration practice.
A 2023 X thread shows asking ChatGPT for a table of composition variables, then turning each row into an image prompt.
In a 2023 X thread, Nick St. Pierre lays out an ordered prompt structure, camera angle first, for interior and architecture images.
Show graphics, campaign films and commercials made with Runway, and animations and 3D scenes built with Rive and Claude.
An animator times and benchmarks one interactive pack-opening animation built four ways in Rive, and finds the CLI versions cost over 3x more per frame.
Harvey's in-house team describes agents building a launch film, hundreds of tailored demo films and a brand library in Figma via MCP.
A Rive educator messaged Claude Code to build an interactive 3D solar system with the Rive CLI, then moved the sun into his course game.
Runway says the agency spent a year testing tools with 26 people before opening a 30-person AI studio in four cities.
Runway says this 2025 studio uses generative AI on every project, with storyboards drawn by hand and AI used for iteration.
Runway says a Paris studio storyboarded 30 shots with AI, cast a model resembling the generated character, and shot two days on green screen.
Runway says a creative studio animated reference images to fill missing archive footage for Maison Margiela and B-roll for a Madonna film.
Runway says a three-person graphics team made a season of textbook-style animal graphics, keeping producer approval before animating.
Runway says production company Tool made an Anthony Joshua spot in four weeks without the athlete present.
Runway says a production team portrayed eleven displaced-background footballers as children without a shoot, then finished the film on 16mm.
AI site builders against a designer's own builds, Figma Make prototypes, agents building Webflow sites, and an AI-assisted carousel.
An agency generates flavour-by-flavour product and fruit imagery for a CPG client's site, then fixes label details by hand in Photoshop.
A product designer rebuilds SkillUp's multi-brand tokens, grid and components with Claude Code in Figma, and lists what he had to correct.
A Webflow designer says a team of Claude Code agents built a native, client-editable Webflow site in three days, and maps where the Webflow MCP fails.
David Demaree fed Trust's original brand Figmas and every past site version to Claude Design and Claude Code to rebuild bitsandletters.com.
A staff product designer prototyped an automation MVP in Figma Make, hit undefined system questions, ran an OOUX sprint, then re-prompted with its outputs.
Rebekah Read reports giving AI tools her real client brief; she found the output functional but short on brand personality.
A principal experience designer says AI let him explore dozens of visual directions for personalized itineraries in two weeks.
A designer reports producing a five-panel carousel for a fintech client in three hours, with AI exploration and manual refinement in Adobe tools.
How a solo designer and an agency team structure briefs and production around AI tools.
A designer and tutor shows how she uses Claude Code, Claude Cowork, Figma and Adobe tools to design, build and market her own products solo.
Toptal says an anonymous insurer’s five-person marketing team cleared a backlog using a brief-to-delivery workflow and AI for early exploration.
What designers need to know about who owns AI output and whether tools train on client files. Not legal advice.
In November 2025 the court rejected Getty’s secondary copyright claim; Getty has since been given permission to appeal.
The Office's January 2025 report says protection depends on human authorship, including selection, arrangement and modification.
In a June 2024 post Adobe said it has never trained generative AI on customer content and would revise its terms.
A 2023 research demo shows text-to-image output matched to the style of one reference image, from watercolor to 3D renders.
According to Adobe, new releases include Photoshop Harmonize and Generative Upscale and Illustrator snapping and gradient tools.
Figma's help center says content training is on by default for Starter and Professional and off for Organization and Enterprise.
Official product information for AI tools and design apps with AI features, linked from the examples above.
Reported October 2025: Affinity combines vector, photo and layout tools in one free app; Canva AI tools need Canva Premium.
Adobe presents it as motion graphics and visual effects software.
Adobe presents it as a design app for social posts, flyers and videos with generative tools.
Adobe presents it as an AI platform for images, video, audio and designs using Adobe and partner models.
Adobe presents it as a vector app for logos, icons and illustrations with precision and AI tools.
Adobe presents it as professional page layout for print and digital publishing.
Adobe presents it as a photo-editing and graphic design app.
Adobe presents it as apps for creating, texturing and rendering 3D models and materials.
Canva presents itself as a visual suite for social posts, videos and presentations, with AI tools such as background removal and a Brand Kit.
Blackmagic presents DaVinci Resolve as editing, color, effects and audio in one application, in free and paid versions.
Eagle positions itself as a visual asset collection tool with tagging and search.
Figma presents itself as a collaborative canvas where design, code and AI come together.
Framer presents itself as an AI design agent that designs and builds sites on a canvas, with hosting and CMS.
Ideogram presents its models and apps as image generation and editing with reliable typography in images, plus an API.
Leonardo.Ai presents itself as a generative platform for images and video, with editing and upscaling.
n8n presents itself as a workflow-automation platform combining AI with business-process automation for technical teams.
Notion presents itself as an AI workspace for documents, search and automation.
Recraft presents itself as a text-to-image model and design platform for photorealism, vector generation, custom styles and mockups.
Runway now describes itself as building “real-world intelligence”, with products built on top.
Spline presents itself as a 3D design tool in the browser with real-time collaboration for interactive web experiences, with a free plan and paid plans.
Stark presents itself as a suite of integrated accessibility tools for design and development teams.
Vectorizer.AI converts PNG, JPG, GIF, BMP and WebP images to SVG, PDF, EPS and DXF vectors. Previews are free; downloading results needs a subscription.
Webflow presents itself as a web platform for designing, building and optimizing sites, with CMS and hosting.
Zapier presents itself as an automation platform connecting thousands of apps, now with AI agents.
In the examples here, AI produced directions, and the designer made the final logo. Linus Ekenstam vectorized a Midjourney logo and cleaned it up by hand (barbershop logo). Huw Mathias redrew an Ideogram mark point by point in Illustrator (VELA). Brandsupply's demo went through three rounds of refinement (Still Touch).
In the US, the Copyright Office's January 2025 report says wholly AI-generated output isn't protected and prompts alone aren't enough. Human-authored parts, and original selection, arrangement or changes, can be (report summary). It isn't legal advice, and other countries differ (Getty v Stability AI, UK).
Adobe says it doesn't train generative AI on customer content, except work submitted to Adobe Stock (Adobe's statement). On Figma, AI content training is on by default for Starter and Professional teams, and only an admin can turn it off (Figma's settings). Check your plan if a client NDA applies.
Not yet. Each entry summarizes someone else's published account, links the source and labels what kind of account it is, including results reported by vendors. A "4TC Recommends" label will only appear on tools we have tested ourselves.
For the bigger question, read Will AI replace graphic designers?