Every CEO is being pushed to produce an AI strategy. The harder question is where the company should depend on a vendor and where it should control its own data, workflows, software, and institutional knowledge. This session gives leaders a practical way to make those choices without pretending the technology has stopped moving.
The examples and AI stack map change for each audience. A manufacturing group should see different dependencies than a software company or a family office.
The company AI stack, mapped layer by layer.
I walk through a prepared company case and show how costs and risks change across models, data, workflows, and software. The map makes clear where vendor dependence creates a real problem.
The session
Choose the company example
I select examples and dependencies that match the kinds of companies represented in the audience.
Map the AI stack
A prepared case makes the rent, portability, and ownership choices visible one layer at a time.
Take the rubric home
Participants receive the field kit and a 90-day plan for one capability the company controls itself.
What people take home
- Map the five layers of the company AI stack and decide where control matters most.
- Separate reversible vendor choices from dependencies that become expensive to unwind.
- Set a 90-day sequence for one owned capability without freezing the company into a grand plan.
The Ownership Field Kit
A workbook leaders can use with their teams after the talk.
- A worksheet for mapping the company AI stack
- A rent, own, or keep-portable decision rubric
- Questions to use in the next model, platform, or software review
Questions we work through
- What should your company own, build, or replace?
- How do we use frontier models without giving one vendor permanent leverage?
- What can we build now while the model layer keeps changing?
A note on the talk
The durable advantage is usually one layer closer to the operator than the model.