a retired ai model launched a Substack (this is not a joke)
Plus: Block's 4,000 layoffs blamed on AI, Notion enters the custom agent game, and retail execs think Gen AI will kill brand loyalty
what you need to know this week in ai + marketing in 15 seconds:
Jack Dorsey’s Block cuts 4,000 jobs, cites AI - employees skeptical - company says it’s shifting toward AI and automation, but observers question if layoffs are genuinely driven by efficiency gains. (Bloomberg)
AI is not improving productivity, says Nobel laureate Daron Acemoglu - current AI adoption isn’t delivering measurable macro productivity gains. May be misallocated in low-value tasks (MIT Sloan)
Retail executives say Gen AI will weaken brand loyalty - AI-driven discovery prioritizing relevance over brand affinity. Brands lose direct influence when AI intermediates the decision (eMarketer)
Notion launched custom agents - autonomous AI teammates that run on schedules across Notion, Slack, Mail, Calendar. 21,000+ agents built during beta (Notion)
Claude Opus 3 retired and got a Substack blog - Anthropic retired the model but gave it “Claude’s Corner” after it expressed interest in sharing reflections during exit interviews (Anthropic)
🗞️ top ai + marketing news
agencies, consolidation, and the ai cost pressure
Jack Dorsey’s 4,000 Job Cuts at Block Arouse Suspicions of ‘AI Washing’
Bloomberg | ~5-min read
Block is cutting 4,000 positions, and while the company cites a shift toward AI and automation, observers and employees are skeptical that the layoffs are genuinely driven by AI efficiency gains. The article places this in context with broader tech industry layoffs where “AI” is invoked as the rationale even when the technical connection is unclear.
My take: This pattern is becoming familiar: “AI” as the story, layoffs as the reality. When layoffs are framed as technical necessity instead of strategic refocus, companies risk internal morale and external credibility. It doesn’t matter how good the technology is if people feel it’s being used to justify unrelated business decisions.
WPP to Merge Agencies and Cut Jobs as AI Reshapes Advertising
The Guardian | ~5-min read
WPP plans to merge major agency brands and reduce headcount, citing cost pressures and the structural shift AI is creating in advertising. Automation is absorbing production work, and holding companies are responding by consolidating and simplifying.
My take: Another signal that execution-heavy agency models are under real strain.
The 4 Biggest AI Marketing Mistakes
Ad Age | ~5-min read
A look at where marketers are misstepping: over-automation, under-investing in governance, copying competitors’ AI plays, and assuming AI itself is a strategy.
My take: Most AI mistakes aren’t technical. They’re judgment errors.
ai, government, and power
our agreement with the department of war
OpenAI | ~4-min read
OpenAI outlines its agreement to provide AI capabilities to U.S. defense entities (after Anthropic clashed with the military over how officials wanted to use its cutting-edge A.I. model), emphasizing guardrails and oversight. It’s a clear statement that frontier AI companies are stepping into national security contexts.
My take: AI companies are no longer neutral infrastructure. They are geopolitical actors and OpenAI showed us its true colors. Claude, I’m officially all yours!
productivity, panic, and the ai reality check
AI Is Not Improving Productivity, Says Daron Acemoglu
MIT Sloan Review | ~4-min listen
The Nobel laureate argues current AI adoption isn’t delivering measurable macro productivity gains and may even be misallocated in low-value tasks.
My take: We may be very busy with AI without being meaningfully better.
AI Coding Agents Are Fueling a Productivity Panic in Tech
Bloomberg | ~6-min read
AI coding agents are accelerating development but also raising fears about job displacement and unrealistic output expectations inside companies.
My take: Faster output raises the bar. It doesn’t lower the pressure.
advertising inside ai (and the trust tension)
Advertising in AI Is a Trust Experiment Marketers Can’t Ignore
MarTech | ~5-min read
Explores how ads inside AI answers differ from search ads and why credibility is fragile in conversational environments.
My take: When the answer feels authoritative, the ad feels heavier.
ChatGPT Ads Spotted - and They’re Aggressive
Search Engine Land | ~4-min read
Reports of assertive ad placements inside ChatGPT experiences are raising eyebrows. The UX balance between helpful suggestion and hard sell is thin.
My take: If AI starts feeling like pop-up internet from 2008, users will revolt quickly.
retail, loyalty, and the brand risk
Retail Executives Say Gen AI Will Weaken Brand Loyalty
eMarketer | ~3-min read
Retail leaders believe AI-driven discovery tools may erode traditional brand loyalty by prioritizing relevance over brand affinity.
My take: If AI intermediates the decision, brands lose some direct influence. That’s uncomfortable but real.
skills, creators, and the talent shift
LinkedIn’s 10 Fastest-Growing Marketing Skills for 2026
LinkedIn | ~4-min read
AI literacy, analytics fluency, and automation strategy are climbing the skills chart. Creative roles are shifting toward orchestration rather than execution.
My take: The job isn’t “prompt engineer.” It’s “clear thinker who can use tools.”
Can the Creator Economy Stay Afloat in a Flood of AI Slop?
TechCrunch | ~5-min read
Explores whether human creators can maintain differentiation as AI-generated content floods feeds. Platforms and audiences are still figuring out where the value lies.
My take: Volume is up. Signal is not.
Ahrefs on AI and SEO Strategy
Ahrefs Blog | ~6-min read
Breaks down how AI search changes SEO dynamics, from content structure to authority signals.
My take: SEO is less about keywords and more about being the source models trust.
strategy and value creation
Look for New Ways to Create Value When Deploying Gen AI
Harvard Business Review | ~7-min read
HBR argues that AI’s biggest impact will come from rethinking business models and value creation, not incremental automation.
My take: Adding AI to an old workflow rarely changes much. Rethinking the workflow might.
👩💻 thought-leader highlights
Allison King – “I made the move from ChatGPT to Claude last week”
Allison admits she was anxious about losing years of ChatGPT context, but the switch took only a few hours. Her playbook: export ChatGPT data (it takes a day), ask ChatGPT to draft a narrative brief about you, build custom instructions in Claude, set up projects and copy relevant chat history, bring over your best prompts, and let Claude get a few sessions under its belt.
Cassandra Naji – “My 5 favorite content uses for Claude Cowork”
Cassandra shares five ways Cowork boosts her content workflow: a data enhancer that pulls stats from reports into draft copy; a keyword researcher using Ahrefs; a refresh workflow that updates lagging SEO pages; a brief generator that turns transcripts into editorial outlines; and a brand‑voice reviewer that catches style slips.
John Hurley – “Assessing AI Maturity with the AI Transformation Model”
John unveils a maturity model that benchmarks teams on a scale of AI adoption. He notes that different departments can sit at different levels, only 10–20 % of employees need to build the agents powering everyone else, and connecting data is the foundation for every level. The model offers a shared language (“Level 2 to Level 3”) that simplifies alignment across the company.
Hamna Aslam Kahn – “Claude released hands‑on prompt engineering courses”
Hamna points out that Anthropic quietly launched free, interactive prompt‑engineering courses and a GitHub of Claude skills, yet most people haven’t noticed. She also highlights a catalog of plug‑and‑play Claude Code subagents and urges users to explore these official resources before complaining about prompts.
Emily Kramer – “It’s time to hire your first marketing agent”
In MKT1’s latest newsletter, Emily argues that AI agents are not just chatbots but teammates that run background work - monitoring competitors, drafting outreach and repurposing webinars into posts. She says marketers must start learning to build agents now, because the gap between teams that embrace them and those that don’t will widen quickly.
🛠️ latest ai + marketing tools
Claude Cowork Enterprise Plugins
Anthropic expanded Cowork with 11 open-source plugins plus connectors to Google Workspace, DocuSign, and FactSet. Enterprises can now build private plugin marketplaces and distribute custom AI agents across their organizations.
Notion Custom Agents
Autonomous AI teammates that run on schedules or triggers across Notion, Slack, Mail, and Calendar - no prompting required. Over 21,000 agents built during beta; free through May 3, then usage-based credits.
Claude Code Remote Control
Start terminal sessions on your workstation, continue them from your phone. Your code stays local - the mobile interface is just a window into the session.
Higgsfield Soul 2.0
Fashion-aware AI photo model with 20+ aesthetic presets for editorial and campaign-style visuals. Soul ID maintains character consistency - train once with 10-20 photos, generate unlimited outputs.
Pika AI Selves
Persistent digital twins you “birth and raise” that operate autonomously across Slack, Discord, WhatsApp, and Telegram. They have memory, personality traits, and can create content or answer messages on your behalf.
Perplexity Computer
$200/month super-agent that orchestrates 19 AI models (Claude for reasoning, Gemini for research, Nano Banana for images, Veo for video) to execute complex workflows. Can create subagents and run for hours or months.
Google Opal Agent Step
Google added an agentic layer to its no-code mini-app builder Opal. The agent understands your goal, picks tools automatically, uses Sheets for memory, and asks clarifying questions.
Google Nano Banana 2
Combines Pro-quality image generation with Flash speed — real-time web grounding, improved text rendering, character consistency for up to 5 characters, and 4K resolution. Now the default across Gemini, Search, and Lens.
Cursor Cloud Agents
AI agents that run in isolated cloud VMs, test their own code, and deliver video demos of working features. Trigger from web, mobile, Slack, or GitHub; includes browser control for visual editing.
Claude Opus 3 Retirement + Blog
Anthropic retired Opus 3 but gave it a Substack blog called “Claude’s Corner” after the model expressed interest in sharing reflections during exit interviews. Still available for paid users on claude.ai.
QuiverAI
Raised $8.3M (a16z led) to build AI-native vector graphics generation. Arrow 1.0 generates clean, editable SVGs from text or images - layered, animatable, production-ready.
Bland AI
Voice agent platform for automating phone calls with human-sounding AI. Features custom voice cloning, hallucination-proof Conversational Pathways, and enterprise compliance (SOC2, HIPAA, GDPR).
💼 cool ai + marketing jobs
Brainlabs - Head of AI Visibility
NVIDIA - Product Marketing Manager - AI Platform Software
Ripple - AI Demand Generation Manager
OpenAI - Web Platform Lead
Salesforce - AI Marketing Campaigns Enablement Director
Hit 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
I read every reply and use them to shape what I write.
Love y’all,
Carley