Welcome to François’ Highlights!

Just last week, a few CMOs I advise asked me for pointers on marketing with AI, given I spend hours per week trying to keep up with the avalanche of news and new AI workflows. So I decided to curate some tips, news, and 💡 insights. My goal is to inspire and equip CMOs and their teams in this AI transformation. 

Feedback and comments welcome!

A selection of AI tips, news and workflows

Tips for using ChatGPT Deep Research 

A must-use tool. Crazy powerful for product marketing (e.g. market or competitive research), researching new software, identifying key influencers, and so much more.

💡 $200/month for ChatGPT-Pro seems high, but if your team shows you they use it often and well, you need to buy it for them. 

Workflow: Create your own personal newsletter

James Raybould shares his workflow for automating the creation of a daily personal newsletter.

💡 Encourage your Content and Growth teams to create personal newsletters for prospects. Better: build an app where customers create their own. Suggest your own content as a source.  

Tools: The best new models for copy writing

The marketer's best friend for copywriting just got an upgrade: Claude Sonnet 3.7 is out, with “extended thinking”. 

💡 Upgrade to Claude for Teams (or higher) and create projects where you add context (PDFs, Google docs) such as your messaging, persona cards, or voice style guide. It saves us a ton of time, and the team writes on message, in the right style.

💡 Do the same for your execs - upload their posts and ask Claude to mimic their style.

ChatGPT 4.5 is rumored to be the best writing model by OpenAI, supposedly improving by leaps and bounds over 4.o on nuance, originality and storytelling. Available to ChatGPT Pro subscribers and devs. 

The GTM Engineer, essential to leverage AI

The hot new role in Growth is called GTM Engineer, a hybrid between Growth and Revenue Ops. They stitch various tools - including AI and automation - and data together to engage customers. 

💡You need at the very least one on your team. 

My favorite listen on AI this week

Kevin Roos and Casey Newton interviewed Dario Amodei, CEO of Anthropic and maker of Claude. I’m a big fan of his communication style. Dario touches on speed of progress, potential impact on jobs, models and security, and more. 

💡 Share this podcast with your CEO. I hope it will convince them to communicate as authentically.

What the rise of agents means for SaaS Marketing Leaders, for now

“2025 will be the year of Agents”. 

Apparently, by December, we should each manage a small fleet of agents. They will take over our repetitive mundane tasks and some of our "strategic ones". A SuperAgent may even supervise and coordinate these task-level agents for us. 

Many are quick to announce tough times ahead for SaaS, even when augmented with AI. Greg Isenberg notes, provocatively:

The current reality, though, is that fully autonomous agents don't work so well, for now. 

Now, some agentic workflows already do work insanely well... Especially for research (see Deep Research in ChatGPT Pro with the o3 mini reasoning model), coding (see Codeium’s Cascade), and tier 1 customer support or analysis.

It helps to think about it as on this continuum:

• Phase 1 = copilots - humans are in the driver seat

• Phase 2 = directed agents - check in with you at each sensitive step

• Phase 3 = fully autonomous agent - perform end-to-end tasks

• Phase 4 = fleet of coordinated agents - could take over entire job categories

💡 Marketing Leaders need to go all-in on 1 and 2 now and start planning for 3. 

Being legally accountable for the work our software performs for us, we will need the right supervision and touch points before letting agents run unsupervised. Unless, of course, there is little at stake (it’s ok to send a few irrelevant prospecting emails when you send hundreds) or there is a testing phase before deployment, as is the case with software. 

💡Question for marketing leaders:

For your products: What agentic workflows to add to your current software offering? What fully autonomous ones should you start scoping?

For your teams: What agents/agentic workflows should your teams be using now, in a few months? How do you inspire and equip them to build, buy, and experiment with agents?  

Final Words: Start with your AI pal…