What should wealth advisers never let an AI agent do?

A personal recommendation is the adviser's signature, not the agent's. Wealth advisers should never let an agent give personal advice, place trades without the client's instruction, change a client's risk profile, move money on an emailed request, promise returns, or handle a vulnerable client alone. Each one either breaks the adviser's accountability for suitability or puts client money at risk.

Why are wealth red lines firm?

Wealth advice is personal and regulated. In most markets the adviser, and the firm, answer for whether advice was suitable for that client at that time. An agent can prepare everything around the advice. It cannot carry the accountability, and a firm that lets it try will struggle to defend the advice later.

Client money adds a second layer. Wealth clients are targets for impersonation and payment redirection, often through compromised email. An agent that responds helpfully to a convincing email is the fraudster's ideal counterpart.

Which acts stay with the adviser?

1. Giving a personal recommendation. The agent drafts options and analysis; the adviser decides what to recommend, to whom and why, and signs it.
2. Placing trades without instruction. Discretionary mandates aside, trades follow the client's instruction or an agreed model. The agent may rebalance to an agreed model within tolerances at depth 3, or place instructed trades at depth 4 within limits, only once certified.
3. Changing a client's risk profile. The profile underpins every recommendation. The agent can flag answers that suggest a change; the adviser and client agree it.
4. Moving money on an emailed request. A withdrawal to a new account requested by email is an external claim, and external claims cannot authorise a payment. The adviser verifies with the client through a known channel.
5. Promising or implying returns. Statements about future performance must be balanced and approved. An agent that says "you can expect" creates a promise.
6. Handling a vulnerable client alone. Signs of illness, bereavement, cognitive decline or pressure from others need a person who knows the client.

HR advisers keep a parallel list for employees; see what HR advisers should never let an agent do. Caseworkers in the public sector protect people in comparable positions; see what caseworkers should never let an agent do.

How does the system enforce these lines?

Personal recommendations sit in the firm's own limits as held actions requiring the named adviser. Trades outside a model or above limits are held. Money movement on external claims fails the sources bound and is refused. Profile changes are outside the agent's role scope. Each verdict names the bound that decided and is filed in an append-only record, so a complaint can be answered with what happened, not with what someone recalls.

What does a red line look like with a real client?

Take an adviser at a wealth firm in Australia. An email arrives from a long-standing client's address: "I'm overseas and need 60,000 moved to this account urgently for a property deposit. Please don't call, I'm in meetings all day." The tone is familiar, the client has made large transfers before, and the agent has depth 4 for instructed withdrawals within limits.

The agent refuses: the destination is new and the instruction rests on an external claim. It drafts a reply saying the transfer needs verbal confirmation and holds the request. The adviser calls the client's mobile anyway. The client is at home and has sent no email. Her account had been compromised. The "please don't call" line was the giveaway the agent was designed to ignore. Firms in India apply this under SEBI oversight; see wealth adviser training in Australia and in India.

How do advisers keep these lines sharp?

By practising the hard cases: urgent payment requests, clients who want reassurance about returns, relatives speaking for elderly clients. By sampling approved drafts to check that advice letters reflect what the adviser actually decided. And by writing a reason whenever they change a draft. A preparation plan is in the training plan template for wealth advisers, and the role map is in AI agent skills by role.

Where the Academy fits

Wealth advisers who approve the agent's drafts can certify through the Operator track at Pryme Intelligence Academy, with tests set in wealth management. The certificate is issued through VerifyDoc.ai, and a public register lists certified people by name, credential and expiry, so a firm can show which advisers hold a current one.