AI for Advisors newsletter
Is your AI tool smart enough for real financial planning?
Back in 2023, when advisors first began testing AI for financial planning, the tools showed a critical weakness: They fumbled basic personal finance questions. They sounded confident—but got the math wrong and the logic muddled.
Two years later, everything’s changed. With the release of new reasoning-focused models, we re-ran those same tests. This time, ChatGPT got it right—both the numbers and the thinking. (We’ll dive into this in the coming weeks.)
It’s a sharp reminder: AI performance is evolving—and the differences between models now matter more than ever. Knowing which model to use is becoming a critical, practical skill for advisors.
Why model choice matters
AI isn’t one-size-fits-all and neither are the models powering tools like ChatGPT, Copilot, and others. Some are optimized for depth and careful reasoning. Others are built for fast, fluent replies.
Knowing which model to use, and when to use it, is becoming just as fundamental as choosing Excel over PowerPoint.
As these tools become embedded in your workflow, it’s essential to understand a key distinction: Some models are optimized for reasoning and depth (ChatGPT’s o3), while others are tuned for speed and responsiveness (ChatGPT’s 4o).
Knowing which to use isn’t technical trivia but a skill that improves the quality of your work and saves time.
Advising people on investing and planning requires judgment, prioritization, and cross-disciplinary thinking. That’s why model choice matters.
Here is the key distinction between ChatGPT’s 4o and o3 models.
- 4o is your fast talker—quick, responsive, and built for everyday conversations. It’s optimized for speed and excels at fast-turn, conversational tasks.
- o3 is your deep thinker—slower but built for complex reasoning. It shines in logic, nuance, and multi-step analysis. Ideal when depth matters more than speed.
- Note: You’re not alone if you think the naming conventions of the models are confusing, you’re not alone!
When it comes to knowing which one to choose for any prompt, it isn’t about performance specs. It’s about selecting the right assistant for the job. Ask yourself who you’d assign the task to in your office:
- Think of 4o as your communications assistant.
- Think of o3 as your senior analyst.
Match the model to the job
One smart way to guide model choice is to think in terms of roles—both yours and the AI’s.
If you’re analyzing complex portfolios, drafting strategy memos, or preparing compliance-related content, you’re wearing a technical hat—and your AI assistant should, too. That’s where o3 models shine.
If you’re preparing a meeting outline or brainstorming a blog post, 4o is likely the better fit.
In ChatGPT, you choose your model in the top dropdown. If you don’t select manually, the default model is likely 4o. You can switch anytime—even mid-task.
| Table 1: Model Matches: What to Use, When |
| Task |
Use o3 |
Use 4o |
| Analyze estate documents |
✔ |
✘ |
| Rewrite a client email |
✘ |
✔ |
| Summarize Roth strategy |
✔ |
✘ |
| Suggest blog titles |
✘ |
✔ |
| Draft investment memo |
✔ |
✘ |
| Prepare meeting agenda |
✘ |
✔ |
| Compare annuity options |
✔ |
✘ |
| Brainstorm LinkedIn posts |
✘ |
✔ |
Use o3 when the task requires depth
You’d want to use o3 because it evaluates things from multiple lenses—macro, portfolio, and regulatory—delivering structured, reliable content. Certain scenarios benefit from slower, more deliberate thinking. That’s o3 territory:
- Estate plan reviews: Uncovering inconsistencies and long-term implications
- Retirement income sequencing: Weighing tax brackets, cash flow, and drawdown order
- Tax tradeoff modeling: Analyzing conversions, deferrals, and charitable strategies
- Compliance-facing content: Balancing clarity with regulatory precision
Prompt example:
“Act as a compliance-focused investment strategist. Write a memo on how rising interest rates affect income-focused portfolios, with attention to suitability and disclosure.”
Use 4o when speed and fluency matters
Many daily tasks are quick-turn and don’t require complex logic. That’s 4os territory.
- Meeting agendas
- Social post ideas
- Simple summaries
- Drafting short emails
- Content outlines
Prompt example:
“Act as a communications assistant. Generate five subject lines for a year-end tax roundup newsletter.”
Try the following example prompts with the different models.
| Table 2: Prompt Examples With Different Models |
| Task |
Prompt |
Best Model |
| Reviewing an estate plan |
“Act as an estate strategist. Analyze this multi-generational trust for tax conflicts and liquidity gaps.” |
o3 |
| Writing a client follow-up |
“Act as a relationship manager. Draft a warm follow-up email confirming today’s action steps.” |
4o |
| Summarizing Roth strategies |
“Act as a financial planner. Create a client-ready summary of Roth conversion tradeoffs across scenarios.” |
o3 |
| Brainstorming blog titles |
“Act as a content strategist. Suggest five titles for a tax-planning seminar.” |
4o |
Add context to improve precision
Remember, regardless of the model you use, AI responds better when it knows the “why.” Context is where AI shifts from generic to genuinely helpful and makes output sharper, more relevant, and easier to act on.
Weak prompt with no context:
“Write a memo on Roth conversions.”
Improved prompt with context:
“Act as a retirement planner. A 60-year-old client with $2.5 million in a traditional IRA wants to reduce future RMDs. Write a memo comparing Roth conversion strategies over the next five years. Ask me questions so we get the best results.”
AI fluency is the next professional edge
As AI tools multiply and specialize, knowing which one to use, and when to use it, is becoming just as fundamental as choosing Excel over PowerPoint.
What sets top advisors apart won’t be whether they use AI, but how well they judge when to lean on speed, and when to pause for deeper thinking.
That kind of discernment—matching the right model to the right moment—will only grow more important as AI handles more of the heavy analysis.
Advisors who embrace these tools gain back what truly matters: the time and space to build trust, navigate hard conversations, and deliver the human insight clients can’t get from software.
AI for Advisors newsletter