If you only check one highlight this week:

• Team leader? How to use personal org charts to accelerate your team's AI adoption.

• Advanced AI users: how a CEO automated many routine tasks at his company with AI agents.

• Everybody else: when to use a single chat, vs a Project vs a custom GPT in ChatGPT

-François

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

Someone forwarded you this email? You can subscribe here.

Two simple ways to accelerate your team’s AI adoption AND proficiency

When a few CMOs asked me recently if I could come in and show their team leads how to use AI better, I told them that I could, if they truly insisted, but that there are more effective techniques than a one-time presentation, since:

1. this acceleration requires both a mindset shift and constant learning

2. things move too fast

3. I bet they are already savvy and creative users and workflows on their teams 

So, instead, here is what I tell execs - and team leaders - to do right now (assuming, of course, that they allow their teams to try tools and unlock budget):

1. Publish personal “AI org charts”

To surface advanced users AND keep people accountable:

👉 Make every person on your team publish their own personal org chart listing the AI workflows and/or agents that “work for” them.

Compile that in a central team folder or wiki.

Agents can be as simple as AI workflows, or GPTs they created and use on a regular basis, or automations they have set up with a step that taps an LLM.

You can use this simple two-slide template and ask folks to update once per quarter.

Review these org charts in one-on-ones.

Have one team member present at each team meeting and showcase their best use cases. Everybody should present at least once.

Recognize folks with the most workflows or agents in the top right quadrant of a “strategic value” vs. “time-savings” 2×2 (see 2nd screenshot below). Ask them to help others build theirs.

Here's mine for instance. Email me back if you want me to share how I built one of them in a future edition.

2. Create space. Repeatedly

AI is easy to use. Benefits show up fast. There is a ton of great resources out there. So often, making progress is as easy as blocking time and sharing notes.

As a leader, you have the power to block off collective time - structured or unstructured - dedicated to experimenting and sharing: an offsite, hackathons, a regular section at your team meeting, during your QBRs, etc.

Or all of it.

You can even create offsites exclusively for your AI champions. This rewards them, makes them feel special, and will help them go further with the best folks on the team or the company.

0. Lead by example

I listed this as step 0, because it's not a technique but a prerequisite to show that fast adoption matters: your team should feel embarrassed that you're using AI better than them. They should perceive that, despite being seemingly in back-to-back meetings, you are investing the time to learn.

Most of the CMOs I work with block recurring AI-learning periods in their calendars. They share learnings and outputs with their teams, then challenge them to improve and build on that work.

You don’t need to learn or keep up with every news, model update, or tool. There's just too much (playing with video generation could be a massive time sink for instance). Instead, pick one or two aspects and get good at those.

For instance, how to go:

1. from using deep research about your target audience or your competitors with Perplexity pro (ChatGPT works too) to compile comprehensive context

2. to feeding them into a ChatGPT or Claude project

3. to creating great outputs with your marketing frameworks or templates you include in your prompts or instructions.

Yes, it is that simple. Telling people to “Use AI” is not enough.

My selection of tips, news and workflows

🤖 When to use a one-time chat vs. a project vs. a custom GPT?

This table by Liza Adams is a good summary. If by now you haven't set up projects or custom GPTs, please just spend 20 minutes this weekend to try that out. Massive time savers and very easy to do (To create a custom GPT, select GPTs in the left panel and then “+Create” GPT at the top right of your screen).

Gemini and Claude have similar features. Claude projects (which I love using - see my personal “agent org chart”) bring together the benefits of both ChatGPT Projects and custom GPTs.

👨‍💻 Advanced: Automating the routine tasks of your team

Sam Levan, CEO and co-founder at Madkudu - explained in this lightning talk (slides) how he identified and automated many repetitive tasks with AI agents. That made his Sales and CSM teams a lot more productive.

His framework:

1. List current activities → identify AI involvement level (0%, 20%, 80%, 100%)

2. Team mindset shift: everyone now manages both people and AI agents

3. Architecture: LLM + MCPs (Model Context Protocol) + Standard operating procedures (SOPs) super well documented

He differentiates between these two modes:

1. Interactive (copilot assistance)

2. Automated (no or little human in loop)

The MCPs you can’t ignore according to Sam:

Company’s stack

• CRM / MAP: Salesforce, Hubspot

• Revenue Intel: MadKudu

• Call recordings(Granola, Gong, …)

• Team Knowledge Base (gDoc, Notion, Confluent ..)

Personal stack

• Web browser: BrowserMCP

• Mail / Calendar

• Todos (Todoist, Linear, Asana, text files)

• Personal Knowledge Base

• Youtube Transcript

After the initial phase that identifies which tasks to augment or automate, he highly recommends spending time documenting your SOPs and compiling in context about your company, products, and more.:

• This documentation becomes your competitive asset

• That knowledge transfers well between organizations

• This data investments compound over time

🎯 ChatGPT: project-specific memory

ChatGPT has had memory for a few months, and that's powerful and convenient. It remembers context across your chats (+ your profile and instructions), unless you turn on temporary chat (top right) or tell it to forget something.

I like telling it to save something specific to its memory, like a style guide or context about a product.

But sometimes, you only want context to be applied to one given project. OpenAI started allowing just that:

When should you turn this on?

I turned to GPT-5 and selected these use cases from its answer:

• Per-client or per-brand workspaces — Why: keeps each client’s voice, assets, and approvals separate so nothing “leaks” into other work.

• Product launch hub (one per launch) — Why: holds roadmap, messaging, FAQs, and technical details.

• Account-based campaign (one per target account or segment) — Why: stores org charts, pain points, relevant case studies & quotes, and custom cadences.

• Crisis communications kit — Why: anchors on approved statements, timelines, and Q&A so responses stay consistent

📝 Monitor complex topics by asking ChatGPT to give you scheduled reports

A tip by Ethan Mollick:

Final Words

François | LinkedIn 

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