your inbox is full of robots

plus: why AI should be an exoskeleton not a coworker, Meta's $2B Manus acquisition hits Ads Manager, and consumers are doubting AI content.

Carley Lake

Feb 24, 2026

what you need to know this week in ai + marketing in 15 seconds:

AI is upending marketing on two fronts - how people discover products (conversational AI replacing search) and how companies execute internally (automation collapsing time/effort). Most orgs are behind on both (HBR)
When everyone can execute fast, strategy gets exposed - AI eliminates execution constraints, so advantage moves upstream to judgment, taste, and prioritization (MarTech)
The path to purchase just got dramatically shorter - consumers deciding faster when AI surfaces answers directly. Less browsing, fewer tabs, more compressed journeys (MarTech)
Email is now a machine-to-machine sport - AI agents summarize, filter, draft, evaluate messages before humans see them. We’re writing for filters and assistant layers now (CMSWire)
Meta’s $2B+ Manus acquisition now in Ads Manager - handles report building, audience research, campaign analysis through conversational prompts. Available to all advertisers (Meta)
Google Lyria 3 in Gemini generates music - 30-second tracks with vocals and lyrics from text/image prompts. All tracks SynthID watermarked, won’t mimic specific artists (Google)
“Stop thinking of AI as a coworker. It’s an exoskeleton” - companies treating AI as autonomous are disappointed. Those using it to amplify human capability see transformative results (Ben Gregory)
Consumers doubt AI content, raising brand trust stakes - growing skepticism toward AI-generated content. You don’t get credit for using AI, only punished for using it poorly (eMarketer)

Thanks for reading! This post is public so feel free to share it.

ai rewriting how marketing works

AI Is Upending Marketing on Two Fronts

Harvard Business Review | ~8-min read
HBR argues AI is reshaping marketing in two big ways: how people discover products (conversational AI replacing traditional search flows) and how companies execute marketing internally (automation collapsing time and effort).
My take: This feels like the right framing. It’s not just a tool shift. It’s a distribution shift and an operations shift happening at the same time.

How AI Eliminates Marketing’s Execution Constraints

MarTech | ~5-min read
Execution used to be the bottleneck. Now overhead is shrinking. When AI absorbs drafting, production, and coordination tasks, advantage moves upstream to judgment, taste, and prioritization.
My take: When everyone can execute fast, strategy gets exposed!! (yup, that’s it!)

The Path to Purchase Just Got Dramatically Shorter

MarTech | ~4-min read
Data shows consumers are deciding faster, especially when AI surfaces answers directly. Less browsing. Fewer tabs. More compressed journeys.
My take: If discovery collapses into one or two AI answers, brand strength matters more than ever. You don’t get ten impressions anymore.

Email Marketing Is Now a Machine-to-Machine Sport

CMSWire | ~6-min read
Email isn’t just humans writing to humans. AI agents summarize, filter, draft, and even evaluate messages before a person sees them. Optimization now includes how systems interpret your message.
My take: We’re no longer writing just for inboxes. We’re writing for filters, summaries, and assistant layers. Ugh. Just when we need more human connection…

measurement, science, and who controls the data

Stanford HAI and AWS Launch Marketing Science Lab Focused on AI Measurement

EdTech Innovation Hub | ~4-min read
Stanford HAI and AWS are building a lab focused on causal inference and AI-powered marketing measurement.
My take: This is the least flashy AI work and arguably the most important.

Consumers Doubt AI Content, Raising Brand Trust Stakes

eMarketer | ~3-min read
Survey data shows growing skepticism toward AI-generated content. Consumers can’t always identify it, but they’re wary of it. Trust drops when AI feels deceptive or lazy.
My take: You don’t get credit for using AI. You only get punished for using it poorly.

Anthropic releases the AI Fluency Index
Anthropic | ~4-minute read
Anthropic published a new research index designed to measure “AI fluency” - the degree to which people understand, interact with, and confidently reason about AI outputs.
My take: This feels like a useful shift in thinking. Most AI adoption metrics are about speed or output volume. Fluency flips the lens to human understanding, which is exactly what marketing teams struggle with right now.

culture, creators, and brand identity

Chatbots Are Becoming Influencers for Brands

The New York Times | ~7-min read
Brands are experimenting with AI personas and chatbot-driven “influencers.” Some are virtual brand reps. Some are full characters. The lines between marketing, entertainment, and automation are blurring.
My take: Virtual influencers still feel gimmicky to me….no matter how much tech has developed…

Generative AI Ads: What Works and What Backfires

Ad Age | ~5-min read
A discussion on which AI-generated creative executions are landing and which are sparking backlash. Themes: authenticity, taste, and restraint.
My take: Audiences aren’t anti-AI. They’re anti-sloppy.

power moves and leadership

OpenAI Hires Charles Porch

Vanity Fair | ~4-min read
OpenAI brings in senior brand talent with deep Instagram experience. This signals something clear: distribution and brand storytelling are now core to AI companies, not secondary.
My take: The AI labs are maturing into full media companies.

Accenture AI Orders Leave Senior Staff Without Promotions

CNBC | ~4-min read
As AI reshapes consulting economics, some senior roles are losing leverage. Automation changes who creates value internally.
My take: AI doesn’t just change marketing strategy. It changes career ladders.

👩‍💻 thought-leader highlights

Ben Gregory – “Stop Thinking of AI as a Coworker. It’s an Exoskeleton.” (Blog)
Ben argues that companies treating AI as an autonomous coworker are disappointed, while those using it to amplify human capability see transformative results. He likens AI to industrial exoskeletons that reduce injuries by 83 percent at Ford plants and emphasizes that AI should augment judgment rather than replace it.

Shae O. – “Something bigger than AI is happening” (Substack)
Shae warns that the real crisis isn’t AI but the decades‑long erosion of attention spans, critical thinking and memory. Delegating cognitive load can increase leverage, she writes, but substituting cognitive effort weakens our capacity; the question isn’t whether we use AI but whether we still practice thinking.

Andreas Horn – “Agentic AI adoption is still mostly in software engineering”
Andreas highlights that nearly half of Anthropic’s agentic AI usage comes from software engineering, while healthcare, finance and legal sectors sit below five percent. He sees this not as slow uptake but as an untapped opportunity for broader industry adoption.

Nicole Leffer – “Instantly improve ChatGPT with custom instruction hierarchy”
Nicole shares a fix for inconsistent ChatGPT performance: set a hierarchy for contradictions so that conversation‑specific instructions take precedence, followed by project‑level instructions, then general custom instructions and finally memory. She says this order improves output quality and urges OpenAI to add a built‑in settings interface for it.

Ann Handley – “Something Messy Is Happening: On AI, Panic and Asking Better Questions” (Blog)
Ann counters viral claims that AI will eliminate half of white‑collar jobs, urging readers not to panic. She argues that the true value of professionals lies in judgment, relationships and the ability to ask the right questions - not just speed - and warns against optimizing purely for efficiency at the cost of meaning.

🛠️ latest ai + marketing tools

Lovable SheBuilds (March 8)
The AI app builder Lovable is going completely free on International Women’s Day. Anthropic is sponsoring with $100 in Claude API tokens per participant, plus $250 in Stripe fee credits. 32 in-person build events across 17 countries. No opt-ins - just show up and ship.

Meta Manus AI in Ads Manager
Meta’s $2B+ Manus acquisition is now fully integrated into Ads Manager and available to all advertisers. It handles report building, audience research, and campaign analysis through conversational prompts. Currently focused on analysis and insights - no campaign execution yet.

Kana Intelligence
$15M Seed New AI marketing agents startup from the founders of Krux, which Salesforce acquired for $700M. The platform uses flexible, modular agents for data analysis, audience targeting, media planning, and campaign management. Mayfield led the round.

Google Lyria 3 in Gemini
Gemini can now generate 30-second music tracks with vocals and lyrics from text or image prompts. Available to all users 18+ across Gemini’s 750M+ monthly active users. All tracks are SynthID watermarked and the system won’t mimic specific artists. Also rolling out to YouTube Dream Track globally.

Figma + Claude Code
New integration lets you capture live UI from Claude Code and convert it to fully editable Figma frames. Type “Send this to Figma” and it becomes a real design artifact you can manipulate. The pipeline works both directions - code to canvas and canvas back to code.

Reddit AI Shopping Search (Testing)
Reddit is testing AI-powered product carousels in search results. When users search for products, they’ll see items mentioned in Reddit discussions displayed with pricing, images, and direct retailer links. Currently a limited US test focused on consumer electronics.

Streak AI
The Gmail-native CRM added two AI features. AI Pipeline Creator lets you describe your business and auto-generates stages, columns, and views. AI Autofill pulls from your email timeline or web research to populate CRM fields automatically.

~ Claude Updates ~

Claude Sonnet 4.6
Anthropic’s new default model for Free and Pro plans delivers near-Opus performance at roughly 5x lower cost. It supports a 1M token context window in beta and excels at complex code fixes and searching across large codebases.

Anthropic Wealth Management Plugins
Anthropic released its first wealth management-specific plugins for Claude, giving RIAs, broker-dealers, custodians, and TAMPs a foundation to build private, compliant AI tools powered by their own data. LPL Financial is expanding its partnership to build AI-powered tools for their 30,000+ financial advisors.

Claude Code Desktop
New features in the desktop app let you write, review, and verify code plus monitor your open pull requests without leaving Claude Code.

Fast Mode for Opus 4.6 (Preview)
A 2.5x faster version of Opus 4.6 is available at higher cost per token. Toggle it on with / fast when you need speed for rapid iteration or live debugging.

Agent Teams (Preview)
Claude Code can now parallelize tasks across multiple agents that coordinate peer-to-peer. Instead of one agent working sequentially, you can split work across specialists that each own their piece and talk directly to each other. Note that running multiple agents increases token usage proportionately.