AI is going to change advisory practice over the next five years. Most of what's been written about it is either breathless ("AI will replace advisors!") or defensive ("AI can't replace the human touch"). Both miss the point.
AI is a tool. Like any tool, it's useful where it's useful and dangerous where it isn't. The work is figuring out which is which — and the firms that do that figuring early are the ones that won't be doing busy-work in 2030 while their competitors are doing planning.
Where AI helps
Drafting first passes of follow-up emails. Not the final email. The draft. The version that's 60% right and saves you fifteen minutes of typing. You edit the tone, fix the specifics, and send. The advisor's voice still ships. The advisor's time is saved.
Meeting summaries. A good transcription tool plus an AI summarizer turns a 90-minute review into a one-page summary with action items in under five minutes. The advisor edits for nuance. The client gets a recap faster than they used to. The CRM gets cleaner notes than it used to.
Compliance pre-review. Run any outbound copy through an AI prompt trained on your firm's compliance standards before it goes to the CCO. Catches the obvious flags. Doesn't replace the CCO. Does cut the review cycle in half.
CRM hygiene. Inconsistent contact records. Missing fields. Duplicate entries. AI is unreasonably good at finding these patterns and proposing cleanups. You approve the changes. The CRM gets better.
Research and prep. Before a prospect meeting, an AI pass on the prospect's company, industry, recent news, and likely concerns gets the advisor 70% of the prep done in three minutes. The advisor walks in sharper.
Where AI kills the relationship
Client communication that's supposed to feel personal. The handwritten birthday card. The condolence note. The "I was thinking about your family this week" check-in. If a client ever figures out you outsourced these to a model, the trust is gone. Don't.
Judgment calls. The reason a client should or shouldn't sell the company. The reason a beneficiary designation matters more than it looks. The reason the recommended allocation is wrong for this specific family. AI is fast at confident-sounding nonsense in exactly these moments.
"AI-written" content under your name. Posting LinkedIn content that was clearly generated. Sending newsletters that read like ChatGPT. Clients can tell faster than you think. The brand cost outpaces any time savings.
Initial discovery. The first time you sit with a prospect, you should be paying attention to them — not coordinating an AI workflow. Tools help after. Not during.
The rule
One sentence: AI for drafts and back-office. Humans for decisions and direct contact.
Or another way to say it: every output that has the advisor's name on it should have the advisor's hands on it. AI can help get to the draft. The advisor still ships the final.
Where to start
If you haven't built any AI into your practice yet, start with one workflow. Meeting summaries are the easiest first win — almost no downside, immediate time savings, the client benefits too. Get that working. Then add follow-up email drafting. Then CRM cleanup.
The advisors who'll lose to AI aren't the ones whose clients prefer humans. They're the ones who refuse to use AI for the parts of their day where it's strictly better. Twenty hours back a month is twenty more hours for the work that actually matters.