Happy Tuesday, everyone!
Every vendor deck this year has implied the same thing without saying it: fewer people, same output. Somebody actually tried it.
Reuters published an investigation on 26 August into what happened inside Meta. Executives explored slashing many teams by as much as 60%, in two waves, with AI absorbing the work. One human-resources executive projected a cut as big as or bigger than the roughly 25% Meta took between late 2022 and early 2023.
It did not happen. Meta says the scenarios included redeployments as well as layoffs, that the company never intended to lay off 60% of its entire workforce, and that what it actually did was move thousands of employees onto newly established teams.
Read that ending again. The most resourced attempt anyone has made to replace headcount with agents ended as a reorg.
Sixty per cent was a scenario on a slide, not a decision, and the piece says so. But somebody modeled it seriously enough for an HR executive to size the redundancies, and it did not survive contact with the actual work.
Today’s sponsor: Aloudable
Aloudable turns your written content into a finished podcast episode in your own voice, without studio time.
Record one short voice sample. We use it to build your voice profile.
Share a document, article, or link. That is the only input we need.
You receive a finished episode and transcript. Review it, request changes, or approve it. Nothing goes live without your sign-off.
Your team provides the source material. Aloudable turns it into professional podcast episodes in your organization’s voice, handling scripting, voice-led production, audio editing, editorial review, and publishing support.
TALK ON THE MARKET
Meta, in a statement to Reuters, on how it evaluates its people:
"Performance rating and promotion decisions were and are made by people, not AI."
A denial is only issued when the assumption is already loose in the world. That sentence exists because enough people believed the opposite for it to need saying, which tells you where the default expectation has landed.
NUMBERS MATTER
Nine in ten marketers say data readiness is critical for adopting AI, and only 21% consider their own CRM data very well prepared for the tools they use or plan to use.
That is Validity's State of CRM Data Report 2026, reported by martech on 28 August, fielded across 500 marketing professionals in July. Everybody agrees the foundation matters. One in five has one.
Today’s partner highlight:
DAILY PROMPT
Create a comprehensive SWOT analysis for [COMPANY/BRAND NAME] in the [INDUSTRY] sector.
Company Context:
- Company: [COMPANY NAME]
- Industry: [INDUSTRY]
- Target Market: [TARGET AUDIENCE]
- Primary Products/Services: [PRODUCTS/SERVICES]
- Company Size: [SMALL/MEDIUM/ENTERPRISE]
- Geographic Focus: [GEOGRAPHIC REGION]
Please structure the analysis with four clear sections:
## Strengths
Identify 5-7 internal competitive advantages, including brand reputation, unique capabilities, market position, technology, team expertise, or operational efficiencies.
## Weaknesses
List 5-7 internal limitations or areas for improvement, such as resource constraints, skill gaps, market presence gaps, or operational challenges.
## Opportunities
Outline 5-7 external growth possibilities, including market trends, emerging customer needs, technological advances, partnership potential, or market expansion.
## Threats
Identify 5-7 external risks and challenges, including competitive pressure, market shifts, regulatory changes, economic factors, or technological disruption.
For each item, provide a brief explanation (1-2 sentences) of why it matters for marketing strategy. Format the output as a clear, actionable document that a marketing team can immediately use for strategic planning.Use this before the restructuring conversation rather than after it. The honest version asks which capabilities you would rebuild from scratch if they vanished tomorrow, and which ones you would quietly be relieved to lose. Meta ran that exercise with consultants and internal documents. You can run it in an afternoon.
CMO CORNER
Every process you delegate without design becomes a future fire.
That is the whole Meta story in nine words. Agents are good at executing a described process and useless against an undescribed one, and most marketing work has never been written down. It lives in the heads of four people who know which client hates red, which approval is theatre, which deadline is real.
Hand that over and you don't get automation; you get a confident stranger. Write down what the person actually does first. Half of it is judgment nobody documented, and that half is why the 60% stayed on the slide.
TOOL UPGRADE
If agents are absorbing work rather than roles, these are the three places it goes first:
Marblism — best for teams who want six narrow agents doing inbox, content, social and lead work rather than one that claims everything. Try Marblism here →
Jace AI — best for the email volume that quietly eats a marketer's week before any strategic work starts. Try Jace AI here →
CustomGPT — best for building on your own documentation, so the answering job stops landing on whoever replies fastest. Try CustomGPT here →
→ We have 300+ curated, filtered, and categorized AI tools in our directory. Browse to find the best ones that help your marketing team.
LEARN THIS
If you are about to run your own version of this, start with the AI Pilot Program Design Framework. It is about scoping a test so it can fail cheaply and legibly, which is precisely what a two-wave, 60% scenario cannot do. For the leader who has been asked to prove agents work before committing to anything structural.
→ We publish workshops, courses, guides, manuals, and more on AI marketing for free. Browse our learning resources for free.
BRAND SLOP
Moonbug Entertainment makes Cocomelon, Blippi and Little Baby Bum. Toddler television, at enormous scale.
404 Media reported on 27 August that the studio has told its animators to start using AI, with written guardrails covering ideation, scripting, storyboarding, design and animation. The rule at the top: keep a human in the loop, AI is used to assist the artist, not be the artist.
Then the guidelines explain why, and it is not about craft. Only what a human creates can be copyrighted, so the final execution has to involve substantial human creative input. Staff is told not to alter a voice actor's performance without checking with Legal.
Not a failure. A company writing its rules before the incident instead of after it, and being honest that the constraint is legal rather than artistic.
Your own AI policy, if you have one, was probably written for brand safety. Moonbug's was written by someone who read the copyright position. One of those two documents survives a lawyer.