AI for Advisors newsletter
An advisor opens ChatGPT and types, “Write me an article on Roth conversions.” Ten seconds later, they have something clean, organized, and surprisingly decent. That feels productive.
But there is a problem. The article may be readable. It may even be practical. But it often lacks the one thing that makes writing persuasive in the first place: judgment.
It doesn’t sound terrible. It just doesn’t say much. That is the trap many professionals fall into with AI-assisted writing. They produce words faster, but not stronger ideas, focused messages, or more distinctive communication.
And in this business, that matters. Writing is not just about filling space. It is about building trust, clarifying decisions, calming fears, and helping clients understand why your perspective is worth listening to.
The real mistake
Artificial intelligence can help you write faster. But it cannot tell you what is worth saying. When advisors get weak results from the model, they often blame the AI or the prompt. Sometimes that is true. More often, the mistake is starting too early.
Many people open the tool and immediately ask it to generate content:
- Write me an article on Social Security.
- Draft a client email about market volatility.
- Create a LinkedIn post on retirement planning.
That feels efficient. But it produces forgettable output because the machine is being asked to do the part that still belongs to you: deciding what matters.
Before any piece of writing becomes effective, you need to take an inventory of your own judgment.
- What is my informed opinion on this topic?
- What have I seen with clients that many people miss?
- Where do people get stuck, even when the numbers are clear?
- What do I believe that is not obvious from a standard explanation?
This is where the value comes from. Not from explaining a topic, but from interpreting it.
Only after that do the practical decisions come into focus:
- What is the argument?
- Who is it for?
- What misunderstanding am I trying to correct?
- What should the reader think, feel, or do?
Those are not prompt questions. They are judgment questions.
And if you skip this step, the model will not fix it. It will simply produce a clean version of an idea that was never fully formed in the first place.
The core principle
Think first. Prompt second. Edit last.
Many people mix up that sequence. They start by prompting, react to the output, and try to fix it afterward. That leads to generic writing, wasted time, and a quiet outsourcing of judgment.
A better approach is to bring AI in after you have direction.
Your job is not to hand over the blank page. Your job is to show up with a central idea. Once you do that, it has something to work with. It becomes a speed layer and refinement tool built on top of your thinking.
The 4-Stage Advisor AI Writing Workflow
1. Think
Before you use AI, answer three questions:
- What is the message?
- Who is it for?
- What should the reader think, feel, or do?
You do not need a full outline. You do need clarity. Many weak output comes from vague thinking. If you are not precise, the model will not be clear. This is where differentiation lives. Not in clever prompts, but in your ability to notice what is happening with clients and turn it into a message that matters.
Let’s look at that useful internal question of “What am I actually trying to say?” and apply it to Roth conversions as an example. A weak starting place is: “Write me an article about Roth conversions.” That produces boiler-plate output because there is no point of view, no audience, no context.
A stronger starting take is: “I want to help pre-retirees understand why Roth conversion decisions often stall emotionally, even when the math looks favorable.”
Now you have an angle, and you are not asking for information. You are asking AI to help express an idea.
2. Prompt
Once you have a clear idea, you can use AI to generate a draft, build structure, and move faster. But the distinction still matters: You are not asking it to invent your thinking. You are asking it to express your thinking.
That requires context and context is the fuel of strong prompts. So, a stronger starting place on the Roth topic would look like this:
Prompt:
I want to write an article for pre-retirees and recently retired clients about why Roth conversion decisions often stall, even after they’ve been shown the tax benefits in their plan. In meetings, many clients understand the numbers but hesitate to act. They dislike writing a large tax check, worry about making an irreversible mistake, feel uneasy about giving up liquidity, or freeze when it is time to decide. I want the piece to acknowledge that hesitation, explain the emotional side in plain English, and help readers move to a thoughtful next step. Ask me a few questions, one at a time, to improve the piece.
This is a stronger prompt because it gives the model something specific to work with: a real audience, a genuine source of friction, and a pointed observation drawn from client experience.
3. Refine
This is often the highest-leverage stage. Many advisors do not need more words. They need sharper, tighter, more specific writing. This is where AI works best as an editorial assistant.
Once you have a draft, run a structured refinement prompt chain. Do not ask it to “make it better.” Ask it to evaluate your draft and then improve it in stages.
Prompt 1: Diagnose the draft
Review the draft below as an experienced copy editor. Identify weak sections. Point out where the writing is unclear, bloated, repetitive, vague, or poorly structured. Do not rewrite yet. Explain what is not working and why.
Prompt 2: Rewrite the weakest sections
Take the weak sections and rewrite them one by one. Make them tighter, focused, and more concrete. Preserve meaning and tone. Show original and revised versions side by side.
Prompt 3: Tighten and smooth the full draft
Now review the full revised draft and improve it further. Cut unnecessary words, reduce repetition, simplify jargon, and replace bland language with more specific phrasing. Improve flow, transitions, rhythm, and readability while preserving the voice. Used in sequence, these prompts force the AI to evaluate before it rewrites.
4. Finalize
This is the step you cannot skip. Even a strong AI-assisted draft needs a human pass, especially in financial services, where tone, trust, and compliance matter.
Before publishing, ask yourself:
- Does this sound like me?
- Would I say this to a client?
- Is the tone right?
- Does anything feel flat or off?
If the answer is no, it is not ready. That is often the clearest sign that AI has polished the language without strengthening the idea. When a draft sounds smooth but forgettable, go back to the beginning and ask: What am I trying to say? That is usually where better writing starts.