Happy Saturday from rainy Boston where I’ve been on a college campus tour with my daughter, wishing I would go back to school and enjoy these incredible campuses.

If you only check one highlight this week:

• if you market SaaS: understand MCP + its potential implications on your business

• others: watch Marketing Against the Grain’s tips for using ChatGPT 4o image gen as an entire creative studio

- François

🛠️ Model Context Protocol (MCP) won the short-lived AI-to-tools protocol race. Agents will rise faster now

Skip to the next section if you don’t like the underlying tech stuff, but I'm covering MCP again today because:

1. OpenAI and Google both embraced the standard introduced by Anthropic

2. Devs love it (see tweet for examples with Figma, Whatsapp, and Github) and will soon be easier to use

3. It could lead to disintermediation and commoditization of “UX-centric” products

4. Just like it's important to understand the concept of API, it's important to understand MCP

So what is MCP?

MCP is a standard for AI to talk to apps. Think of it as a universal connector for AI that sits between AI models and the (APIs of) the tools and data we use.

It gives AI:

1. tools - i.e. the ability to take actions beyond generating text, images or code

2. access to data - i.e. deeper context for better recommendations or decisions

When AI taps our business or personal tools, it can perform more tasks for us. That’s the concept of agents. For instance, we can connect a desktop AI client (such as Claude) to MCP servers (not trivial yet) and use it to: scrape websites, use Zapier automations, send emails or WhatsApps, use Figma, all from that client.  

It will likely pose significant challenges to UX-centric SaaS products since their products could lose direct contact with customers when users interact with their services trough ChatGPT, Claude, Gemini or other 3rd party tools that access these services in the backend, on their behalf.

This will challenge products that monetize when we view ads (e.g. Yelp) or buy upgrades (airlines, freemium). It should favor API-first players (e.g. Twilio) if they embrace MCP fast and provide LLM-optimized docs.  

The MCP ecosystem is still in its early days, so there are still questions about discovery, security, authentication, and more.

Want to learn more? Here’s an AI-generated intro podcast built on sources and posts I collected. Or request access to the Notebook LM to chat with that knowledge base and/or learn through the mindmap that tool from Google created.

My selection of tips, news and workflows

🛠️ How to use AI as our entire creative studio

Block some time to watch this recent episode of Marketing Against the Grain (Audio - YouTube - 30 min) is really insightful. I recommend watching vs listening to see their prompts.

It details how to conceptualize, create, and deploy marketing assets with ChatGPT, including its new 4o image gen capabilities (you can do similar things with Gemini).

Kip (Hubspot CMO) and Kieran (SVP Marketing) show, among other things, how to:

• Use deep research to create comprehensive brand style guides based on sites and brands you love

• Build complete campaign assets with mood boards and multiple formats

• Set up multi-stage workflows: research → prompt crafting → generation

It’s so rich, a tl;dr wouldn’t do it justice.

🛠️ User of Google Workspace? Tips to get more out of Gemini

You may, as I do, have many files in Google Workspace. That makes Gemini, the AI “closest to your knowledge”. Good for us, Gemini keeps getting better and better, as the Neuron calls out. It’s great value if you don’t want to use open source LLMs or find ChatGPT Pro ($200) or Claude Max ($100 or $200) prohibitive.

And on top of it, Google just announced “AgentSpace” to create our own agents and connect them to our favorite online tools.

» Yes, Google’s been catching up. Given how much of our data it has and how much we trust it with it, it’s now a formidable force in AI.

📊 The 7 traits of best-in-class AI Marketing teams

According to Jasper’s State of AI and marketing 2025 report (nice landing page btw), best-in-class AI marketing teams share seven key traits:

1. proper AI use case documentation

2. leadership buy-in

3. workflow integration

4. experimentation mindset 

5. have a marketing AI council 

6. offer advanced training

7. use marketing-specific AI tools, not just general-purpose models (of course…)

» The best marketing teams move beyond good prompting; they set up strong, repeatable, and automated workflows.

My clients asked for me for a survey to:

1. Identify their AI champions

2. Set a baseline for current AI usage and impact

3. Identify workflows ripe for AI augmentation

📨 Email me (reply with your company name) if you want a version for your team.  

🏫 CEOs now expect AI literacy from their teams

There was much buzz this week about the leaked memo from Shopify's CEO where he lays out expectations of his employees regarding AI.

He’s not the only CEO demanding that. If you’re reading this, you probably agree: we should all augment our work with AI, especially if it increases job satisfaction… and security!

Here are steps to make your team an AI-first team:

1. Challenge and equip your team. Unlock budgets for AI and training. Free up time to learn and experiment. Cut other resources a little

2. Enable your primary LLM, as you enable Sales, with the right knowledge. Appoint a “Marketing Intelligence manager” for that

3. Hire for AI-first skills and mindset. Ask to see candidates’ workflows

4. Assess and reward your team for effective use of AI

You’ll find more details in highlights #2.

Final Words

Thanks for sharing these highlights with busy marketing execs around you.🙏 

François | LinkedIn 

I'm a CMO, advisor, and "CMO Wingman." Yes, that's a thing :-). Ask my clients: in this AI era, CMOs need a strategic and supportive advisor more than ever. I’m former CMO at Twilio, Augment Code, Apollo GraphQL, Decibel, Udacity and Head of Marketing for LinkedIn Talent Solutions.