What is prompt engineering?

Definition

Prompt engineering is the practice of writing prompts carefully: giving the AI a role, context, a clear task, the format you want and examples. It also means testing a prompt on several real cases and adjusting it until the output is reliably good.

For most advisors it isn't a technical skill. It's the same as writing a clear brief for a new assistant.

Why it matters for advisors

A well-built prompt saves review time. If your follow-up email prompt already says "no performance promises, plain English, under 200 words, list action items with owners," fewer drafts need fixing.

It's also a control. Firms that standardize prompts for client-facing tasks get more consistent output, which makes supervision easier. Approved prompts can live alongside your AI policy.

Save prompts that work. A shared library of tested prompts for meeting summaries, client emails and document reviews saves time and makes AI output more consistent across the firm. Our prompt library is a starting point.

Example

An operations lead tests a prompt for turning meeting notes into CRM tasks on ten past meetings, fixes where it missed deadlines, and adds the final version to the firm's prompt library.

Watch out for

Clever prompts can't fix a tool that isn't approved for client data. Start with the right tool and plan, then improve the prompt.