Because AI is such a hot topic and the frontier of its use seems to be continually changing, I’m periodically sending out an update about how I’m using AI today. (You can compare to my use cases from 9/2023.) I would love to hear back from you about how you’re using AI, also.

Some of you will find my usage fairly basic, for others, it might feel advanced. I’m just one data point as we all try to navigate this new tool.

Things that are working well.

I find myself using AI most often as a brainstorming thought partner. When I’m at the beginning stages of creating something new, whether I’m looking for a recipe or a new business strategy, starting a conversation with the AI helps me to clarify the parameters of what I’m looking for—and also gives me some baseline feedback to narrow my thought processes.

Several custom GPTs are giving me very consistent results.

• I’ve trained a CustomGPT to create proposals. My proposals have a similar format but are customized for each client. To achieve this, I set up a custom GPT with an example proposal and demonstrated to it which parts to keep the same and which to change in each proposal. If I give it my meeting notes, it does an outstanding job of creating a proposal that requires just a few edits.
• I also have a CustomGPT to help me write LinkedIn posts to publicize podcasts that I’ve been a guest on. Here again, I provide the transcript of the podcast recording and ask it to identify a hook that will be engaging on LinkedIn. Once we agree on the hook, the AI writes the body of the LinkedIn post. This has taken quite a bit of fine-tuning, but it’s also now working pretty well.

Tools

In terms of tools, I switch back and forth quite a bit between ChatGPT and Gemini. I prefer Gemini for writing and ChatGPT for thinking. For research, I either use Perplexity or the Deep Research tool in Gemini.

If you haven’t tried Gemini’s Deep Research (OpenAI has a similar product), it’s worth giving it a spin.

For example, I took a bunch of my ideal clients and found their LinkedIn pages, then copied 10-20 of their posts into a document. I then asked Deep Research to create a client persona from that information. After 20 minutes or so, it had researched each client’s background, found other things each had posted online, and made a very detailed report on the amalgamated persona of those example clients. I’m now using that client persona to evaluate different marketing copy or promotional ideas to see how that persona would react to those things. It’s provided me with some valuable food for thought.

I’ve seen other folks create competitive analysis reports, troubleshooting manuals, and buyers’ guides using it with solid results.

Experiments I’m running.

I started experimenting with AI for coaching. I created a custom GPT loaded with information from several coaches that I’ve worked with in the past and from some other coaches that I admire—and asked it to base its coaching on those ideas.

This also took some fine-tuning. Out of the box, the coach doesn’t ask a lot of questions; it feeds me interventions pretty quickly, without a lot of insight. But with some fine-tuning, I have gotten some interesting results from my AI coach. Where it’s been most valuable is in helping me to clarify my messaging and get over some of my hang-ups about how I talk about my work. It’s also given me some feedback that reinforced what I was hearing from the humans in my life, spurring me to take action!

My next frontier

Between the AI coaching and other strategic conversations where I’m using AI as a brainstorming partner, I’m starting to get concerned about the amount of private information that I’m using in these tools.

While I know the provider’s privacy policies tell me that they won’t use this for marketing or training, I was on Facebook in 2007, and I remember how things changed there. So I’m cautious about giving big companies too much of my personal information. Their incentives are not aligned with my interests!

I’m starting to explore some AI tools that have a zero data retention (ZDR) policy. Neither the prompts nor the AI replies are stored on the provider’s servers. These tools don’t have all the bells and whistles that OpenAI or Gemini have. So, I’m thinking about segregating out some of my more private conversations, or work that includes sensitive client information, into a tool like Open Router instead of using the big frontier models for those conversations.

What’s working for you with AI right now?

Hit reply and tell me, what’s your best use case? What have you tried that failed utterly? Let’s learn together!