Multi-year AI cost and exit planning
On This Lesson 4 topics
This lesson is planned. The outline below lists the topics it will cover. The full lesson has not been published yet.
In this lesson
compare total cost over several years including usage growth and switching.
Planned topics
- Usage-based pricing grows with adoption: model cost as conversations times tokens, with growth scenarios and peaks; verify prices before recording.
- Add the other costs: retrieval, evaluation, monitoring, integration, people, support, and for self-hosting, hardware and operations.
- Price changes and model deprecations happen: include a switching cost and a periodic re-evaluation.
- Exit plan: keep prompts, evaluation sets, and retrieved data portable; use an abstraction layer; know how data is returned or deleted.
- Show best, expected, and worst scenarios so finance sees the range.
Draft outline details
Demonstration
a three-year table for Larkspur: year, conversations per month, cost per conversation, platform and people cost, switching reserve, for each of the three approaches under expected growth.
Check your understanding
Why include switching cost in the model?
Show answer
Models and prices change, so leaving a vendor must be affordable and planned.
Practice
Apply this lesson with the module's lab, exercises and quiz.
Marking a lesson complete is optional and you can undo it any time. Your progress stays in this browser.