Start with a small task your team already knows

Matt Ragone of Range describes a practical way to build useful AI skills: start with work you already understand well enough to check.

In his AI Engineer talk, Ragone reported roughly 24 skills across the company after six to eight months, including 17 in financial planning. Planners without coding backgrounds built on written procedures their team already kept in Notion and Google Docs.

A process you can borrow

1. Pick a frequent, bounded task whose output you can judge.
2. Teach it with finished examples, including cases you had to correct.
3. Include clarifying questions and clear limits.
4. Test it on typical and unusual cases before sharing it.
5. Assign an owner and reviewer to keep it current.

For Range’s planning workflows, the output remains a draft until a certified financial planner reviews it before delivery to a client. That distinction matters: making a draft faster does not remove responsibility for checking it.