I watched two AI personas argue about a product while one played a skeptical buyer. Then AI told me what the rep did wrong.
Took less than five minutes to build. No code. ChatGPT, Gemini, and Claude can do this with some differences on how the output is rendered and shared.
This is what I’m calling a thinking system. It’s AI using AI. You set up the scenario, and AI-powered personas interact with each other, each response a live AI call. You’re not driving. You’re watching AI talk to AI while you learn.
The economics of this kind of training didn’t make sense before. Now anyone can build it.
There are two modes of building with AI right now:
Knowledge tools – you drive the conversation. AI assesses, advises, creates.
Thinking systems – AI invokes AI. You observe, learn, get coached.
Both enable work that wasn’t possible before. The difference is your role.
Key insights:
Knowledge tools put you in the driver’s seat. Thinking systems are AI using AI. You set up the scenario, AI-powered personas interact, each response generated live. You observe and learn.
Strategy docs, playbooks, and research reports can become systems that train, coach, and challenge your team. Documents your team created but rarely uses are full of expertise waiting to be activated.
You don’t need to be technical. Describe what you want, upload a document, and let AI build the system. Refine from there.
Using this approach can build a significant advantage as AI capabilities expand. Learning to think in terms of “what systems could I design?” is the skill that keeps paying off.
Thank you to Jennifer Pockell Dimas (CMO at Telarus) and Mary Gilbert (Kerford) at (Founder of InfiniteEdge) for sharing their perspectives on what this shift means for GTM teams.
See the full breakdown, a working demo you can try, and practical steps to build your own in the newsletter below.
I also shared a Buying Committee Simulator example. Super interesting to see who’s bought in and and who’s not after a meeting with the sales rep.
I created an 6-min AI video explainer and 12-min AI podcast versions (links in comments) of the newsletter using NotebookLM to cater to different learning styles. I reviewed these AI outputs personally for accuracy and to ensure responsible use. Check them out. This tech is advancing dramatically to help with learning.
Would love to hear your thoughts. If you found this helpful, please subscribe and share.