Timeliness reminder / Time-bound snapshot: This article is a trend distillation of29 design and AI tools industry news items collected by Will (dated roughly late 2025 — early 2026).Specific features, version numbers, and who launched first will all become outdated; what is truly durable are the following 6 trend signals (which direction things are heading). Read it for the "direction", don't memorize the "feature list". This is a point-in-time read of ~late-2025–early-2026; treat the direction, not the feature list, as the durable takeaway.
Reading guide: Each trend = one-sentence signal + several pieces of supporting news + implications for PMs/designers (Why it matters to you). It does not restate every news item one by one.
First, a quick primer on industry jargon (used directly later):
- Liquid Glass: A new visual language launched by Apple in 2025, where UI elements resemble translucent frosted glass, letting content behind show through and refracting with light.
- Vibe coding: A new development approach that does not write traditional code, but instead "gestures at a real running product page while AI syncs changes back to code".
- Generative UI: Interfaces are no longer fixed pages drawn and finalized by hand, but are assembled/drawn on the spot by AI based on immediate needs.
- Everything App / Super App: An app that tries to pack "watching, chatting, shopping, paying, playing" all into one place, so users never need to leave it.
Trend 1 · The "glass-mimic" design wave sparked by Apple Liquid Glass (one platform sets the tone, the whole industry follows)
Signal: With iOS 26's Liquid Glass, Apple has established "translucent glass texture" as the new mainstream aesthetic,and the entire toolchain aligned collectively within a few months—this is a replay of the classic script where "a platform vendor defines the visual language and third-party tools rush to adapt" (paralleling Material Design / iOS flat design in their day).
Supporting news:
- Apple iOS 26 Liquid Glass + Beta3 rollback: The first version of glass was too transparent, making Control Center icons hard to see; Beta3 increased frosting and lowered transparency, pulling back a step between "looking good" and "being legible".
- Figma launches Glass effects + Apple official Liquid Glass UI Kit: Figma rolled out glass effects with adjustable light angle/refraction/depth of field/chromatic aberration, and Apple simultaneously released an official UI Kit, letting designers create iOS 26-compliant interfaces in one click.
- (Related) PANTONE 2026 Color of the Year Cloud Dancer: A warm "perceptually friendly white", shifting from "high-contrast pure white" to "soft and non-glary", following the same broad direction of "reducing visual burden" as glass/frosted textures.
Implications for PMs/designers · Why it matters:
- Aesthetics are "dealt" by platforms, not freely chosen by you. As soon as Apple moves, Figma/PANTONE/countless apps have to follow, or they look "outdated". When building AI products, UI skins should anticipate such waves—don't rebuild a style that is about to become obsolete right at the peak of the trend.
- The Beta3 move is a key lesson: No matter how trendy a visual style is,once it harms readability/usability users will push back. Glass feel is "decoration"; whether information can be seen at a glance is the "foundation". Before adding flashy effects to your product, first ask "can users still recognize buttons under strong light/dynamic backgrounds?". For visual hierarchy principles see Google Material Design 3 Design System (Core Claims and Rules for Dynamic Color/Type Scale/Tonal Elevation/Shape/Adaptive Layout/Motion/Accessibility) (Material's elevation/contrast rules follow the same underlying logic).
- Glass texture works better on large screens / AR (such as Vision Pro), but use it cautiously in information-dense backoffices/forms.
Trend 2 · The AI design tool arms race among Figma/Adobe/Canva/Google (tools shifting from "canvas" to "generation engine")
Signal: All creative tool giants are competing for the same positioning—"one-stop AI creative workspace". Shared moves: ① Turn high-frequency time-consuming operations (cutout/expand image/remove background/typesetting) into built-in one-click AI capabilities; ② Upgrade from "helping you draw" to "helping you generate finished products that remain editable". The essence of design tools is migrating from editor → generation engine.
Supporting news:
- Adobe Acrobat Studio Adobe Acrobat Studio: Combines Acrobat (strict document control) + Express (lightweight design) + Firefly (AI generation) into one workspace, turning PDFs from "read-only final drafts" into "conversable, manipulable knowledge containers (PDF Spaces)". One selling point is that AI-generated content comes with clickable source citations (playing the enterprise compliance card).
- Canva Creative OS / Design Model Canva editable design AI: Touted as "the world's first editable design AI model"—AI directly outputs real design files with layers/typesetting/brand structure (rather than a flat dead image), plus @Canva in comments as an inline assistant, and Canva Grow taking over the full chain of "inspiration → batch generation → run Meta ads → view data".
- Figma new-generation AI image editing: Three high-frequency tasks—expand image/cut out object/remove object—are unified into a contextual toolbar that "appears when you point at an image", so non-designers can do things that previously only designers could do.
- Google Mixboard Google Mixboard: A Gemini-powered "AI concept board" that generates visual direction directly from a single sentence (benchmarked against Pinterest + AI), compressing "find images → align style → organize" into one step.
Implications for PMs/designers · Why it matters:
- "Editable" is the real watershed in this round. Early AI output an image that was useless if you couldn't modify it; outputs like Canva's that are structured files you can keep editing only then does it truly enter the production workflow. When evaluating any AI creative tool, the first question to ask is: "Can its output be further edited, and can it be accumulated into reusable assets?"
- AI pushes the cost of "making one version" close to zero → value shifts from "knowing how to make" to "knowing how to judge/choose". The scarcity of designers/PMs shifts from hands-on ability to aesthetic judgment + maintaining brand consistency + choosing the right direction.
- Compliance/traceability is becoming a B2B selling point (Adobe's source citations). If your AI product is entering enterprise/government/legal scenarios, "can what AI says be traced to a source" will be a hard procurement requirement, worth designing for in advance.
- Giants are all building "one-stop" experiences, which means the living space for single-point small tools is being squeezed—to build an AI product you either go deep enough vertically, or have a data/ecosystem moat.
Trend 3 · AI-generated interfaces + Vibe coding rewriting the "design-development" boundary (UI shifting from "dead pages drawn by humans" to "AI-generated on the fly")
Signal: This is the most fundamental of the 6 trends, and the one most likely to disrupt industry role divisions. Two threads are happening simultaneously: ① vibe coding Vibe coding—developers modify directly on real pages and AI syncs back to code, the wall between design and engineering is collapsing; ②Generative UI—the interface itself is assembled/drawn on the spot by AI, rather than all pages being pre-designed.
Supporting news:
- Cursor's new development approach: Merges "see interface, modify interface, write code" into one continuous experience—developers select elements, drag layouts, adjust font size and color (still following design system rules) on real running pages, and Cursor syncs to the code layer. The IDE is shifting from "a place to write code" to "a place to directly build the product", and the traditional division of labor between design and engineering is deliberately weakened.
- A2UI (open protocol, Agent-driven UI): A2UI:AI does not write code or render directly, but only outputs a segment of "structured interface description" (what components I want, how components relate), and the client renders using its own already trusted component library. This resolves the contradiction of "wanting AI to participate in interfaces, but not letting untrusted remote AI arbitrarily modify the frontend". Flow: AI describes the interface → says "ready to render" → user actions are passed back to AI → AI decides how the interface changes next.Interface = something that can be safely passed between Agents just like data.
- Flipbook (real-time interface generation for the AI era, 5.4M views): Uses the LTX video model to generate interfaces frame-by-frame in real time (target 1080P/24fps, WebSocket streaming), with no DOM/frontend code/fixed components; essentially "live video + real-time interaction"—as soon as the window changes, the entire picture reconstructs itself.
Implications for PMs/designers · Why it matters:
- The traditional pipeline of "design mockup → asset slicing → implementation" is being bypassed. Cursor turns "modifying the product" and "modifying code" into the same thing; teams will increasingly collaborate around "the real product itself" rather than around mockups and PRDs. PMs need to rethink:When prototype equals production and a change equals going live, how should requirement reviews/design handoff change?
- A2UI lays the foundation for the "multi-Agent era". In many future tasks, pure chat is too slow (ordering food/booking appointments/filling forms), a form plus a few buttons is actually more efficient—so AI should not only generate text, but also interfaces. But security/brand consistency requirements dictate: let AI propose "requirements" and the client do the "implementation", rather than letting AI spit out HTML directly. PMs building Agent products should treat "when AI should present an interface, and whose component library the interface should render with" as a core design question. See UX Interaction Paradigms in the AI Era (AI embedded in traditional interfaces / Agentive / Conversational / Human-in-the-loop / Hyper-personalization / Multimodal / Trends — seven types total)(Among these, the two paradigms "Agentive UX" and "Human-in-the-loop" are precisely the interaction backbone of this trend).
- Flipbook sits at the more radical end (generating the entire interface in real time): it currently looks more like a show-off demo, but it reveals the ceiling —interfaces no longer need to have "fixed components". Practical takeaway for PMs: in the short term, the controllable approach is the compromise represented by A2UI — "AI selects components, client renders them" — rather than full-screen direct AI output.
Trend 4 · Platformization of creative tools + design references "coming alive" (from tools to platforms, from static to workflows)
Signals: Creative/design infrastructure is upgrading in two directions: ① Individual tools are growing into platforms/operating systems(Canva calls itself a Creative OS); ② Design references are upgrading from "static screenshots" to "playable real flows", while mainstream software is collectively redesigning their visual systems to align with "AI-era aesthetics".
Supporting news:
- Mobbin offers "playable animated flow demos": instead of only providing static interface screenshots, it shows real operation flows (button feedback / page transitions / loading states / transition pacing).Upgrading from a "screenshot collection" to an "interaction reference library", which creates higher barriers and is harder to copy.
- Major Sketch redesign (Copenhagen): fixed sidebar replacing floating panels, layer Focus mode, adaptive toolbar,one-click local AI background removal(Apple on-device ML), automatic line wrapping for Stacks. The core is not adding features, but clearing away years of accumulated interface burden in one go and moving closer to macOS's new design language.
- Microsoft Office Fluent 3D icon redesign: evolving from flat design to a fluid, spatial 3D language; Microsoft's subtext is "AI is not a tool, but a gentle, flowing companion", so the icons adopt a rounded, fluid style and form a scalable icon matrix.
- YouTube cross-platform UI redesign: player information aligned to the left for greater immersion, comments changed to a threaded structure, 20+ animation effects for the like button.
- App Store Web version: rebuilt from jump-style static pages into a full Web App; the "Today" content section is fully brought to the Web, and product pages are media-oriented —making App Store content indexable by search engines and shareable across social platforms.
Implications for PMs/designers · Why it matters:
- "Reference libraries" are also becoming AI-powered/dynamic. When you look for design references for your team, static images are no longer sufficient (they cannot show loading feedback, state changes, transition pacing — the details that truly determine experience quality).To evaluate whether a flow is good, you need to see whether it is smooth when it "moves".
- Big tech companies are collectively redesigning their visual systems, with a shared subtext of "presenting AI as a gentle, flowing companion" (Microsoft 3D icons / Apple glass / Pantone soft white) —the visual language of AI products is shifting from "cold tech feel" to "approachable companionship". When building AI companion/tool products, this is an emotional tone worth referencing.
- Web presence = new traffic entry point(App Store Web makes content searchable/shareable). When building products, don't only focus on in-app experiences;Web surfaces that can be indexed by search engines and spread socially are an underrated growth lever.
Trend 5 · The "retention + monetization" arms race among super apps and social platforms (keeping users and money inside the ecosystem)
Signals: Social/content platforms are fighting two battles: ①Fighting for retention— every design choice serves "don't let users leave" (competing with TikTok, building an Everything App); ②Digging for monetization— building an in-app economic closed loop around virtual goods, creation, and tipping.
Supporting news:
- X tests new link format + Everything App: when clicking external links, web pages open in an overlay, with the original post still visible, so users can continue liking and commenting without exiting. The open play is Musk's "super app" strategy —reducing external-link churn is a prerequisite for building a super app.
- Instagram's three-pronged response to TikTok: testing making Reels the main home screen, splitting the bottom-nav Following feed into three streams: All/Friends/Latest, with full-screen swipe switching; also launching Repost to main Feed + Instagram Map social map(more restrained than Snap Map), to shore up the weakness of "social-graph-driven distribution".
- Meta Vibes Meta pure AI video feed: TikTok-like pure AI video feed, turning "AI video generation" from a tool into a social feed of "consumption + secondary creation", with the generated data feeding back into model training (a moat).
- Telegram gift economy (two articles): gift favorites / limited-edition gifts / gift valuation and resale market / user ratings based on Stars transactions / paid submissions (Stars + Toncoin escrow + automatic refunds) / BotFather Mini App —overall evolving toward "Web3-like virtual assets + content crowdsourcing middleware", moving from a chat tool toward a super app.
- Spotify Mix: letting Premium users add transitions/EQ/cover stickers to playlists —upgrading "listening to music" into "creating", turning listeners into curators/remixers, boosting stickiness and UGC.
- TikTok adds voice/images to direct messages + minor nude photo detection + Minor nude photo detection: closing the social loop while playing the compliance card.
Implications for PMs/designers · Why it matters:
- "Reducing exits" is the core theme of retention design. X's overlay links and Instagram's swipeable feed switching are essentially reducing task interruption and shortening paths. When building products, every action that makes users "leave the current scene" (external links / page jumps / secondary confirmations) is a retention leak point worth examining one by one.
- Monetization is expanding from "ads" to "in-app economies"(Telegram gifts/paid submissions, Spotify creation, Meta AI-UGC). The common playbook:turn consumer goods into collectible/show-off-able/resellable "assets", then add scarcity mechanisms (limited editions) and social display slots— emotional value drives repeat purchases. For content/social/companion products seeking monetization, this "virtual assets + collection showcase + scarcity" combo is worth breaking down.
- "Lowering the creation barrier so ordinary users can participate" is a platform consensus(Spotify Mix, Meta Vibes Remix, Telegram submissions). Packaging "professional capabilities" into "one-click/secondary creation" is the standard tactic for activating UGC.
Trend 6 · AI reshaping search/translation/maps/localization (from "giving you information" to "helping you make decisions")
Signals: The task model of utility products is changing —from "I give you information, you judge for yourself" → "I understand your situation and directly prepare you to make a decision". At the same time, AI is making cross-language content circulation nearly zero-cost.
Supporting news:
- Google Maps' biggest redesign in 17 years: real-time 3D maps (generated by Gemini) + cinematic camera positioned ahead of the car(to give "anticipation time") + "Ask Maps" natural-language directions (you can ask even when your phone is low on battery or you don't want to wait in line). Design philosophy:"keep your eyes on the road" — reduce screen dependence, frontload information. One-sentence summary:Moving from Search UI to Decision UI(from "retrieving information" to "contextualized decision support").
- Google Web Guide Google Web Guide: Gemini intelligently chunks search results by topic (running multiple searches simultaneously via query expansion), reducing information overload for vague queries — a navigational guide via "structured pre-classification".
- Google Translate transforms into a Duolingo competitor: Gemini generates personalized language lessons based on your level/scenario (tap-to-listen / AI conversation / real-time bilingual conversations in 70+ languages), and is also testing mimicking your own voice to speak foreign languages. Upgrading "translation" into "language practice", entering the EdTech market.
- Meta AI multilingual translation + lip sync: Reels are automatically translated and lip-synced, preserving original tone; creators can "create once and reach globally", with up to 20 language audio tracks.AI directly flattens the "language gap", the biggest barrier to content globalization.
- (Related) Gmail shopping logistics tracking view + smarter promotions page: aggregating scattered emails into a unified view, ranking promotions by engagement — a small entry point for AI in "organizing information and reducing cognitive load".
Implications for PMs/designers · Why it matters:
- "Search UI → Decision UI" is the paradigm shift utility products should copy most. Users often don't want "a pile of results for me to pick from", but "recommendations with reasons + directly actionable next steps" (book/save/share/navigate). When designing AI features,merging the high-friction chain of "research → filtering → planning → execution" into one continuous experience is more valuable than simply "searching more accurately". This aligns with the evolution direction of "conversational → intent-driven interfaces" in UX Interaction Paradigms in the AI Era (AI embedded in traditional interfaces / Agentive / Conversational / Human-in-the-loop / Hyper-personalization / Multimodal / Trends — seven types total).
- "Frontloading information / giving anticipation time" is a universal technique for reducing cognitive load(Maps camera positioned ahead, turn-by-turn broken into small decisions). Any design that forces users to "react on the spot and make urgent decisions" is a source of burden —letting users see the next step in advance applies to almost all flows beyond navigation.
- Content localization will soon no longer be a barrier(Meta/Google translation + lip sync + voice mimicry). For PMs building overseas products: language is no longer a moat;think in reverse — can your content leverage this wave of "zero-cost multilingual" dividends to create once and reach globally?
One-page cheat sheet · TL;DR (for those in a hurry)
| Trend | One-sentence signal | The one thing PMs should remember most |
|---|---|---|
| 1 · Glassmorphism wave | Apple sets the tone, the whole industry follows | Flashy effects must not hurt readability (Beta3 rolled back) |
| 2 · AI design tool arms race | Everyone is racing for a "one-stop AI creative workspace" | "Editable outputs" are the watershed; value shifts to judgment |
| 3 · AI-generated interfaces | UI moves from human-drawn to AI-generated on the fly | AI defines requirements, client renders (A2UI compromise) |
| 4 · Tool platformization | References come alive + collective visual redesigns | Visuals shift from cold tech to gentle companionship |
| 5 · Super app arms race | Locking in users + in-app economic closed loop | "Reduce exits" + "virtual assets + scarcity" monetization |
| 6 · AI reshaping utility tools | Search UI → Decision UI | Give recommendations with reasons + actionable next steps, merge high-friction chains |
A reminder:This is a point-in-time snapshot for 2025–2026. Features will iterate, and the order of who launches first will change, but directions such as "tools becoming generation engines", "the collapse of the design-development boundary", "Decision UI", "reducing exits", and "zero-cost localization"directions will most likely outlast the news itself.
See also
- UX Interaction Paradigms in the AI Era (AI embedded in traditional interfaces / Agentive / Conversational / Human-in-the-loop / Hyper-personalization / Multimodal / Trends — seven types total) — Full panorama of AI interaction paradigms (conversational / agentive / human-in-the-loop / multimodal…), where the interaction backbone of Trend 3 and Trend 6 in this article is expanded
- Google Material Design 3 Design System (Dynamic Color / Type Scale / Tonal Elevation / Shape / Adaptiveness / Motion / Accessibility — core propositions and rules) — Design language system (elevation / contrast / hierarchy); shares the same underlying usability logic as the glass texture in Trend 1
- Airbnb Design Language and Historical Releases (2021–2025 quarterly release timeline + DLS design system + trust/belonging design) — Brand-level design language case, comparable to Trend 4's "big tech visual redesigns"
Sources and related materials
- UX Interaction Paradigms in the AI Era (Seven Trends: AI Embedded in Traditional Interfaces / Agentive / Conversational / Human–AI Collaboration / Hyper-Personalization / Multimodal / Trends)
- Google Material Design 3 Design System (Core Propositions and Rules: Dynamic Color / Type Scale / Tonal Elevation / Shape / Adaptive Layout / Motion / Accessibility)