3 minute read

In my last post , I shared a detailed plan for swapping assets between a taxable and a tax-deferred account to optimize tax efficiency. I used an LLM to build the plan itself, and I was impressed with how thorough and actionable it was. This experience sparked a few conversations with others about how LLMs might disrupt the financial services industry. Personally, I think this is one area where these tools are going to have a huge impact — helping people refine strategies, make better decisions, and process complex information more efficiently.

Today, I ran another quick experiment. This time, I focused on using the LLM to distill and summarize financial strategies by interacting with it in a conversational, iterative way. What I’m sharing here isn’t just the output — which is useful — but also the process. The key takeaway is that you can work with these tools using plain, simple language without overthinking how to phrase things.

Here’s what I did, with all the prompts embedded so you can see exactly how I used the tool and how you might apply it for your own needs.

Step 1: Asking a Broad Question

I began by asking the LLM for a list of financial strategies I might not know about. My exact prompt was:

“I’m relatively financially savvy, but I recently learned how ETFs can avoid passing on capital gains due to in-kind redemptions, and it surprised me. Can you list 10 other financial strategies or concepts like this that I might not know about but would find valuable?”

The response was packed with useful strategies, such as asset location, tax-loss harvesting, charitable donations of appreciated stock, and the 529-to-Roth conversion strategy. While this initial list was helpful, it needed more depth and actionable details to make the ideas practical.

Step 2: Refining and Adding Examples

To make the strategies more actionable, I refined my prompt and explicitly asked for clear examples while keeping the explanations concise. Here’s exactly what I asked:

“Can you take the list you just gave me, prioritize the ideas by ease of implementation, and include clear examples in less than ten sentences for each item? I want it to be simple, direct, and understandable by someone with moderate to high financial savviness.”

The LLM refined the list, prioritized strategies based on ease of implementation, and added examples that made the concepts easier to understand. For instance, it detailed how donating appreciated stock directly to a charity avoids capital gains taxes while maximizing your deduction. It also explained how you can convert unused funds from a 529 plan into a Roth IRA for the beneficiary, giving them a head start on retirement savings. These structured examples made the concepts clear and actionable.

Step 3: Formatting for Clarity

Once I had the refined content, I wanted it formatted for easy sharing. I provided specific instructions to the LLM for this as well:

“Can you format this list in a way that’s easy to copy into a document or print as a PDF? Include clear headings, concise explanations, and examples that fit within ten sentences for each item.”

The result was a well-organized list with headings, concise descriptions, and actionable examples. It was polished and ready to share without needing additional edits. Here’s the final document

Why This Matters

Using AI tools like LLMs is incredibly easy and effective for refining strategies, summarizing complex ideas, and generating actionable insights. The real strength of these tools lies in how quickly you can iterate to reach high-quality outcomes in a short amount of time. You don’t need perfect prompts or technical expertise — just treat the interaction like a conversation.

What makes this even more accessible is the ability to use voice input, which eliminates friction and allows you to experiment with ideas anytime, anywhere. These tools are powerful for anyone willing to dig in and see how they can work for their specific needs. The best way to learn is to get started — ask questions, refine your approach, and see where it takes you.

Updated: