Google stopped being a search company
Welcome to this week's AI + Marketing Weekly! Plus: checkout happens in chat now, California's AI layoff order, and the consumer trust gap widening.
~The AI + Marketing Weekly~
what you need to know this week in ai + marketing (15-second version):
Google I/O 2026 focused entirely on making Gemini the operating layer across every product - Gemini 3.5 Flash (fastest frontier model), Gemini Omni (any input/output), Gemini Spark (24/7 agent), Universal Cart (shopping across Search, YouTube, Gmail). Google wants to be an intelligence system, not just an OS. (Google I/O)
Google Marketing Live introduced Ask Advisor, a Gemini agent connecting Ads, Analytics, Merchant Center, and Marketing Platform - plus AI Mode ad formats for conversational search. AI Search is the only search now. (Google Marketing Live)
Universal Cart lets shoppers buy multiple products inside chat and transfer to retailer sites - Google is building rails where purchase happens in chat and never touches your site. This decides who owns the customer at sale. (Marketing Brew)
Newsom signed first-of-its-kind executive order on AI workforce disruption - mandates studies on severance and employment insurance. Signed a day after Meta’s AI layoffs. (Office of the Governor)
97% of marketing leaders use AI daily, but 78% of consumers prefer ads made by people - 87% say best advertising needs human touch. People aren’t anti-AI, they’re anti-soulless. (Canva)
AI engines evaluate brands less on marketing narrative, more on customer experience - your reviews are now your positioning. You can’t write your way into an AI recommendation if the experience is mediocre. (MarTech)
Aitana Lopez has 400,000 Instagram followers, major brand deals, and she’s AI - one in three Gen Z buys based on AI influencer recommendations. The backlash is about feeling deceived, not the technology. (Fast Company)
Only 49% of martech tools are actively used - “workslop” happens when teams are pushed for volume with no quality control. Knowing which 20% is wrong still falls to a human. (MarTech)
Figma launched a native AI agent on its canvas - generate, remix, and edit designs through natural language. Revenue hit $333M in Q1, up 46%. (Figma)
🗞️ top ai + marketing news
california moves on ai jobs
Newsom signs first-of-its-kind executive order on AI workforce disruption
Office of the Governor of California | ~4-min read
Newsom directed state agencies to prepare workers for AI-driven job loss, signed a day after Meta’s big AI layoffs. The order mandates studies on severance, employment insurance, and worker ownership, plus recommendations within 180 days on updating mass-layoff notice law. For now it’s diagnostic, not enforceable.
Why it matters: A study is not a protection, and workers losing jobs this year won’t feel anything change. But naming AI layoffs as a distinct policy problem is the kind of move other states copy.
google goes all-in on agentic
Google’s new AI Mode ad formats and agentic commerce tools
Marketing Brew | ~5-min read
Google introduced two AI Mode ad formats with Gemini-generated explainers, both testing in the US. It also launched Ask Advisor, an agent that pulls Merchant Center details to build campaigns and surfaces insights across Google Ads and Analytics. The agents now carry context between each other.
Why it matters: “Our agents talk to one another and carry each other’s context” is Google describing a marketing team made of bots. The human’s job becomes setting direction and catching the misses, a very different skill from running the campaign yourself.
Google Marketing Live 2026: the full announcement collection
Google | ~8-min read
Google framed Gemini as the foundation across Search, YouTube, and commerce. The Universal Commerce Protocol is expanding with a Universal Cart that lets shoppers buy multiple products inside chat and soon transfer items to the retailer’s site. Asset Studio got multimodal Gemini creative generation.
Why it matters: Google is building the rails for a world where the purchase happens inside the chat and never touches your site. The Universal Cart is the line item to watch, because it decides who owns the customer at the moment of sale.
the workslop problem
Marketing teams must own AI, or workslop will take over
MarTech | ~6-min read
Only 49% of martech tools are actively used and just 15% of organizations qualify as high performers. “Workslop,” a flood of low-quality generic output, happens when teams are pushed for volume with no time for quality control. The fix is an AI inventory, a mission statement, and clear handoff lines between teams.
Why it matters: Workslop is the perfect word for AI mandated without anyone owning how it gets used. Volume was never the hard part. Knowing which 20% is wrong still falls to a human, and that’s the job worth protecting.
consumers vs. the ai ad machine
Canva finds consumers want useful AI ads, but value human input
Marketing Tech News | ~4-min read
Canva’s State of Marketing and AI report surveyed 1,415 marketing leaders and 3,547 consumers. 97% of leaders use AI daily, but 78% of consumers would rather see ads made by people and 87% say the best advertising needs a human touch. 74% would feel better if formal AI policies governed its use.
Why it matters: The gap between 99% of marketers leaning in and 78% of people wishing a human made it is the whole story of this year. People aren’t anti-AI, they’re anti-soulless. Disclosure is starting to look like a brand signal worth owning.
The AI backlash is a danger for every brand now
Fast Company | ~5-min read
A Stanford and UC Berkeley poll found fewer than half of Americans want the country to charge ahead with AI. The risk has widened from the AI labs to any brand seen using AI in dubious ways. A Nike post touting Jannik Sinner drew accusations of AI-generated copy, a small sign of the new mood.
Why it matters: AI is an inauthenticity force multiplier. The scary part is you no longer have to use AI badly to get burned, you just have to look like you did.
the new shopping signals
Customer experience outweighs brand in AI-assisted shopping
MarTech | ~6-min read
When AI engines evaluate brands, they focus less on marketing narrative and more on customer experience, building your brand from reviews, comparisons, and forums. The old visibility playbook still matters but isn’t enough. Better CX now produces the signals that decide how often AI recommends you, making it an acquisition lever.
Why it matters: Your reviews are now your positioning, whether you like it or not. You cannot write your way into an AI recommendation if the experience behind your brand is mediocre. This is the most accountable marketing has ever been.
What customers consider when using AI to shop
CX Dive | ~4-min read
A survey of 4,200+ consumers found price is the top factor in AI-assisted shopping, then ratings, then the AI’s top pick. For an unfamiliar brand, two-thirds would consider it but need to check reviews first. AI-discovery shoppers are 2x more likely to abandon at any friction, and 4 in 5 won’t return after a bad experience.
Why it matters: People let AI do the shortlisting, then fall back on the same instincts: do I know this name, what do others say. Patience for friction is razor thin, so a clunky checkout for an AI-referred shopper is a customer you never see again.
ai influencers go mainstream
She has 400,000 Instagram followers and major brand deals. She’s also AI
Fast Company | ~5-min read
Aitana Lopez is a “virtual soul” built by Barcelona agency The Clueless, with a backstory, a cat, and a team of eleven behind her. Despite rising anti-AI sentiment, one in three Gen Z consumers buys based on AI influencer recommendations, and nearly half of college-age Gen Z follow at least one.
Why it matters: The same generation that says it hates AI is buying from synthetic people, which tells you the backlash is about feeling deceived, not the technology. If the persona is fun enough, plenty of people stop caring whether she’s real. That should unsettle every brand still arguing about disclosure.
design, everywhere
Canva’s new Gemini integration makes AI graphic design ubiquitous
Fast Company | ~4-min read
Canva is now integrated with every major AI model: Claude, ChatGPT, Copilot, and Gemini. Users can generate designs from chat context and pull them into Canva to edit, with Brand Kit keeping everything on-brand from the first prompt. Canva serves 265 million monthly users.
Why it matters: Canva’s whole strategy is to live inside whatever tool you already have open, and it’s working. For non-technical marketers, the gap between “AI made me an image” and “I have a brand-ready asset” was always the annoying part. That’s the step Canva just removed.
👩💻 thought-leader highlights
Hamna Aslam Kahn - “Advanced techniques for maximizing Claude AI’s capabilities”
Hamma outlines 11 expert-level strategies including Projects for persistent context, identity files, collaborative mindset treating Claude as thinking partner, writing style matching, extended thinking, meta-prompting, and removing preambles with “No preamble. No filler. Just the answer.” Requires approximately 10 minutes setup for compounding value across future sessions.
Cecilia Stallsmith - “From blank prompt → published URL — no Eng queue, no tickets, no waiting”
Cecilia announces a live 45-minute session on May 28 demonstrating how product marketers can create and publish landing pages without developer assistance using no-code “vibecoding tools.” Removes traditional development bottlenecks, democratizing landing page creation for non-technical marketing professionals.
Alexandra de Lara - “What women are actually using Claude for”
Alexandra explores practical AI adoption by interviewing female entrepreneurs using Claude for competitive research, survey analysis, SEO optimization, website building, expense tracking, and AI “Chief of Staff” for priority management. Core philosophy: delegate repetitive administrative tasks—research summaries, spreadsheets, code snippets—freeing mental energy for strategic thinking rather than abdicating decision-making.
Anna Wojciechowska - “The Learning Curve Nobody Budgets For”
Anna examines counterintuitive productivity dip when adopting AI tools. Identifies four phases: Weeks 1-3 excitement, Weeks 3-6 the “crash” where editing overhead exceeds time savings, Weeks 6-12 learning what helps, Month 3+ genuine gains. Key insight: “Stop using AI for things you’re already fast at.” Two to three months represents realistic minimum before measurable improvements.
🛠️ latest ai + marketing tools
Google I/O 2026: the big picture Google’s biggest I/O in years, focused entirely on making Gemini the operating layer across every product it touches. Key launches for marketers:
Gemini 3.5 Flash — faster than any frontier model, now powering AI Mode and Search
Gemini Omni — any input, any output
Gemini Spark — a 24/7 personal agent running in the background on your phone or laptop
Universal Cart — a cross-app shopping cart working across Search, YouTube, and Gmail
Redesigned Search box — built for longer, conversational queries
Intelligent Eyewear — Android XR-powered glasses from Samsung with Gentle Monster and Warby Parker
Google is done being an operating system and wants to be an intelligence system.
Google Marketing Live 2026: what marketers need to know Google’s annual ads event made one thing clear: Gemini is now the operating system behind Google’s entire ad and commerce stack. Key launches:
Ask Advisor — a single Gemini agent connecting Google Ads, Analytics, Merchant Center, and Google Marketing Platform for unified campaign management
AI Mode ad formats — new ad units designed for conversational search journeys
Universal Cart — rolling out to Search and the Gemini app this summer
Gemini Omni Flash in Asset Studio — text-to-video creative production for ads
Google Antigravity 2.0
Google’s agent-first development platform got a major upgrade at I/O, moving from coding tool to full orchestration platform for building and deploying agents. Now runs on Gemini 3.5 Flash, supports multi-agent coordination, and integrates natively with Google AI Studio and Android. Gemini Spark is built on top of it.
Pomelli agentic capabilities
Google’s local discovery AI, now in Europe, gained new agentic capabilities that let it complete tasks on your behalf rather than just answering questions. Can now book restaurants, send emails to local businesses, and surface options based on your personal Google data.
Figma Agent
Figma launched a native AI agent embedded directly on its collaborative canvas, letting teams generate, remix, and edit designs through natural language without leaving the file. Respects existing design systems, supports multiple simultaneous agents, and integrates with Claude Code and Codex via MCP. Free during beta for Professional, Organization, and Enterprise plan users. Revenue hit $333M in Q1, up 46% year-on-year.
ASOS Stylist in ChatGPT
ASOS launched a ChatGPT app that lets shoppers discover and browse its full product catalogue through conversation, including shoppable video and livestream content. Built with video commerce platform Bambuser, it returns curated edits and styling advice in real time, with a “Shop on ASOS” button to complete purchase. Live for UK and US shoppers now.
Metabind
Build MCP apps once and render them natively across Claude, ChatGPT, and any MCP host, using a SwiftUI-like component system called BindJS. Write in TypeScript; it renders as native SwiftUI on iOS, Jetpack Compose on Android, and React on web without rebuilding for each platform.
OpenAI content provenance
OpenAI is embedding C2PA content credentials across all its image generation tools, so every AI-generated image carries verifiable metadata about how it was made. Pairs with Google’s expanded SynthID watermarking announced at I/O. Relevant for any brand creating AI-generated visuals that need to be trusted.
Codex computer use (locked mode)
OpenAI added locked computer use to Codex, letting agents execute tasks on a fully isolated remote machine without touching your local environment. Available now for Enterprise and Business plan users. Designed for high-security workflows where the agent needs browser or app access but must stay sandboxed.
Stability AI Stable Audio
Stability AI launched Stable Audio as a standalone product for AI music and sound generation. Generate custom tracks, sound effects, and atmospheric audio from text prompts. Available via API for production workflows and as a consumer tool. Relevant for content teams building video, podcasts, or branded experiences that need original audio.
Unframe raises $50M
Unframe, the managed AI delivery platform that takes enterprise AI from pilot to production in days, raised a $50M Series B and crossed $100M in total contract value in 12 months. LLM-agnostic, integrates with any SaaS or database without data leaving the enterprise perimeter, and uses outcome-based pricing so clients don’t pay until results are live. 400% net revenue retention.
Lovable Google connectors
Lovable now connects directly to Gmail, Calendar, Drive, Sheets, Slides, Maps, Gemini Enterprise, and BigQuery, letting builders create full-stack apps on real Google Workspace data with no API setup required.
💼 cool ai + marketing jobs
Perplexity - Demand Generation Lead
Reddit - Creative Technologist, KarmaLab
Snowflake - Product Marketing Lead - AI Solutions
Product.ai - Director of Brand & Content
Anthropic - Head of Anthropic Creative, Creative Studio
Reply anytime and let me know:
Topics you’d like covered in future weeks
Sections you love or could do without
Questions about AI, marketing, or anything else
Love y’all,
Carley